Preface

This document is the user manual for Apusic In-Memory Data Cache (AMDC) V2.0.4 cache core product, detailing the functionality usage, configuration methods, and management operations of AMDC cache core.

Target Audience

This document is intended for AMDC product operations engineers, IT system operations engineers, and development engineers.

Related Documentation

For more information about AMDC V2.0.4 product, please refer to the following AMDC V2.0.4 product manual documentation set:

No.	Manual Document	Description
1	Apusic In-Memory Data Cache V2.0.4 Quick Start Guide	Brief introduction on how to quickly get started with AMDC.
2	Apusic In-Memory Data Cache V2.0.4 Installation Guide	Detailed introduction on how to install AMDC on various operating systems, AMDC service start/stop operations, and product registration process.
3	Apusic In-Memory Data Cache V2.0.4 Cache Core User Manual	Detailed introduction on the usage, configuration, management of AMDC related functions and supporting tools.
4	Apusic In-Memory Data Cache V2.0.4 Console User Manual	Detailed introduction on the usage and operation instructions of AMDC console related functions.
5	Apusic In-Memory Data Cache V2.0.4 Development Guide	Detailed introduction on AMDC client application development based on various programming languages.
6	Apusic In-Memory Data Cache V2.0.4 Migration Guide	Detailed introduction on migrating AMDC historical versions to V2.0.4, and migrating Redis to AMDC.
7	Apusic In-Memory Data Cache V2.0.4 Operations Guide	Detailed introduction on AMDC monitoring, operations, security hardening and other operational instructions.
8	Kingbee Apusic In-Memory Data Cache V2.0.4 Performance Optimization Guide	Detailed introduction on AMDC performance tuning.

Technical Support

AMDC product provides comprehensive technical support services. You can obtain technical support through the following channels:

• Website: www.apusic.com
• Phone: 400-855-5800
• Email: support@apusic.com
• Kingdee Cloud Community: https://vip.kingdee.com/?productId=73&productLineId=14&lang=zh-CN

When requesting technical support, please provide the following information:

1. Your name
2. Company information and contact details
3. Operating system and version
4. Product version number
5. Detailed information about exceptions and errors, including logs and screenshots

Introduction

Apusic In-Memory Data Cache Software (Apusic In-Memory Data Cache, referred to as AMDC) is a fully scenario-applicable, high-throughput, secure distributed cache software that provides secure and reliable caching support for large-scale, high-concurrency, high-availability critical applications. The product is compatible with Redis protocol and persistent data files, allowing simple, fast, and smooth replacement of Redis.

Related Concepts

Term	Meaning	Usage Notes
Standalone	Stores data, handles data read/write, handles data synchronization	AMDC's default mode, standalone mode. In standalone mode, the master node runs in standalone mode.
Master-Slave	Replicates data from master node, handles data read/write, does not participate in data synchronization	Real-time replica of master node data. It actively connects to the specified master node and receives the data stream sent by the master node, thereby maintaining data synchronization with the master node. By default, slave nodes are read-only, and all write operations must be sent to the master node.
Sentinel	Monitors master-slave nodes, handles master node failover, handles master-slave node synchronization	Sentinel mode. When the master node goes down, it automatically switches to a slave node. When a slave node goes down, it automatically switches to master node.
Cluster	A computer service system composed of multiple independent computers	In this document, master-slave/sentinel/cluster modes can all be considered clusters. Cluster mode specifically refers to the mode with multiple master nodes for data sharding storage.

Feature List

Feature	Description	Additional Notes
Multi-data Type Cache	Provides data cache types: string, list, hash, set, sorted set, geo, hyperloglog, stream.	Supports Redis protocol compatible data structures, can directly replace Redis without modifying application code
Pub/Sub	Implements publish/subscribe functionality, enhancing system capabilities	Supports channel subscription, pattern subscription, can be used for message queues, event notifications, etc.
ACL Permission Control	Provides secure access mechanism for services, supporting fine-grained access permission control.	Role-based permission management, supports command-level and keyspace-level access control
Memory Data Eviction Policy	Provides multiple data eviction policies to meet various data eviction requirements and improve memory utilization.	Includes LRU, LFU, TTL and other eviction algorithms, can be flexibly selected according to business characteristics
Persistence	Improves availability for cache core, prevents data loss during crashes.	Supports RDB snapshot and AOF log persistence methods, ensuring data security
Lua Script Support	Supports using Lua scripts to operate cache core.	Supports atomic operations for complex logic, improves processing efficiency
Multi-threading Mode	Supports multi-threaded concurrent request processing, improving system throughput.	Uses I/O multiplexing technology, fully utilizing multi-core CPU performance
Master-Slave Mode	Supports master-slave backup.	Implements data redundancy, supports read-write separation, improves system availability
Sentinel Mode	Provides node monitoring, automatic failover, failure notification, configuration propagation for master-slave mode.	Implements high availability architecture, automatically detects failures and switches services
Cluster Mode	Supports elastic scaling (memory expansion/shrinking), and also has failover functionality.	Data sharding storage, supports online expansion and shrinking, meeting big data volume scenario requirements
Monitoring & Operations	Performance Monitoring	Provides detailed performance metric monitoring, including QPS, response time, memory usage, etc.
	Slow Query Log	Records commands with execution time exceeding threshold, facilitating performance tuning
	Statistics	Obtain server running status and statistical data through INFO command
High Availability	Automatic Failure Recovery	Automatically detects node health status, implements automatic failover
	Data Consistency	Ensures data consistency between master and slave nodes
	Service Discovery	Supports automatic discovery of cluster nodes, simplifying configuration management
Security Features	Access Control	Supports password authentication, SSL/TLS encrypted transmission
	Audit Log	Records user operation logs, meeting compliance requirements
Extension Features	Transaction Support	Supports MULTI/EXEC and other transaction commands
	Pipeline Operations	Supports batch command execution, reducing network round-trip time
	Key Expiration Management	Supports TTL, EXPIRE and other key lifecycle management functions
Development Support	Client Compatibility	Fully compatible with Redis client libraries (Jedis, Lettuce, etc.)
	Protocol Compatibility	Supports RESP protocol, can seamlessly replace Redis
	Toolchain Support	Provides CLI tools, benchmark tools, etc.

Product Installation

For AMDC cache core installation, please refer to Apusic In-Memory Data Cache V2.0.4 Installation Guide

Redis Configuration Compatibility

Starting from AMDC v2.0.2, it is compatible with Redis configuration files. AMDC will automatically convert the content in redis.conf to AMDC configuration content.

Usage: ./amdc-server redis.conf (specify Redis configuration file)

Usage Introduction

Cache Core Usage Guide

Distributed caching is the core capability of AMDC and the center of the entire product. Other functions are built on the data caching business. AMDC stores data directly in memory, uses multi-threaded read-write separation to achieve efficient storage, meets different types of data storage requirements, enables rapid development, and reduces type conversion. It supports multiple data eviction policies to reasonably utilize memory space.

Operation Commands

Command	Description
ACL LOAD	Reload ACL from configured ACL file
ACL SAVE	Save current ACL rules to configured ACL file
ACL LIST	List current ACL rules in ACL configuration file format
ACL USERS	List all configured ACL rule usernames
ACL GETUSER username	Get rules for specific ACL user
ACL SETUSER username [rule [rule ...]]	Modify or create rules for specific ACL user
ACL DELUSER username [username ...]	Delete specified ACL user and related rules
ACL CAT [categoryname]	List ACL categories or commands within category
ACL GENPASS [bits]	Generate pseudo-random secure password for ACL user
ACL WHOAMI	Return the name of the user associated with the current connection
ACL LOG [count or RESET]	List latest ACL security events
ACL HELP	Display helpful text about different subcommands
APPEND key value	Append value to key
ASKING	Sent by cluster client after -ask redirection
AUTH [username] password	Authenticate to the server
BGREWRITEAOF	Asynchronously rewrite append-only file
BGSAVE [SCHEDULE]	Asynchronously save dataset to disk
BITCOUNT key [start end [BYTE|BIT]]	Count set bits in string
BITFIELD key [GET encoding offset] [SET encoding offset value] [INCRBY encoding offset increment] [OVERFLOW WRAP|SAT|FAIL]	Perform arbitrary bitfield integer operations on strings
BITFIELD_RO key GET encoding offset	Read-only variant of arbitrary bitfield integer operations on strings
BITOP operation destkey key [key ...]	Perform bitwise operations between strings
BITPOS key bit [start [end [BYTE|BIT]]]	Find first bit set or clear in string
BLPOP key [key ...] timeout	Block, pop and return first element
BRPOP key [key ...] timeout	Block, pop and return last element
BRPOPLPUSH source target timeout	Pop last element from list and push to head of another list; blocks if list is empty until timeout or element available
BLMOVE source destination FROM-TOP|FROM-BOTTOM TO-TOP|TO-BOTTOM timeout	Block, pop and delete first or last element from source list, push to first or last position of destination list
LMPOP numkeys key [key ...] LEFT|RIGHT [COUNT count]	Pop one or more elements from first non-empty list key in provided key list
BLMPOP timeout numkeys key [key ...] LEFT|RIGHT [COUNT count]	Block, pop one or more elements from first non-empty list key; or block until another client pushes or timeout
BZPOPMIN key [key ...] timeout	Block, pop and return member with lowest score from one or more sorted sets, or block until another client pushes or timeout
BZPOPMAX key [key ...] timeout	Block, pop and return member with highest score from one or more sorted sets, or block until another client pushes or timeout
CLIENT CACHING YES|NO	Indicate whether server should track keys in next request
CLIENT ID	Return client ID of current connection
CLIENT INFO	Return information about current client connection
CLIENT KILL [ip:port] [ID client-id] [TYPE normal|master|slave|pubsub] [USER username] [ADDR ip:port] [LADDR ip:port] [SKIPME yes/no]	Kill client connection
CLIENT LIST [TYPE normal|master|replica|pubsub] [ID client-id [client-id ...]]	Get list of client connections
CLIENT GETNAME	Get current connection name
CLIENT GETREDIR	Get tracking notification redirect client ID (if any)
CLIENT UNPAUSE	Resume processing of paused clients
CLIENT PAUSE timeout [WRITE|ALL]	Stop client command processing for specified duration
CLIENT REPLY ON|OFF|SKIP	Indicate whether server should reply to commands
CLIENT SETNAME connection-name	Set current connection name
CLIENT TRACKING ON|OFF [REDIRECT client-id] [PREFIX prefix [PREFIX prefix ...]] [BCAST] [OPTIN] [OPTOUT] [NOLOOP]	Enable or disable server-assisted client cache support
CLIENT TRACKINGINFO	Return information about server-assisted client cache for current connection
CLIENT UNBLOCK client-id [TIMEOUT|ERROR]	Unblock client blocked on blocking command from different connection
COMMAND	Get array of Redis command details
COMMAND COUNT	Get total number of Redis commands
COMMAND GETKEYS	Extract keys given full Redis command
COMMAND INFO command-name [command-name ...]	Get array of specific Redis command details
CONFIG GET parameter [parameter ...]	Get configuration parameter values
CONFIG REWRITE	Rewrite configuration file with in-memory configuration
CONFIG SET parameter value [parameter value ...]	Set configuration parameter to given value
CONFIG RESETSTAT	Reset statistics returned by INFO
COPY source destination [DB destination-db] [REPLACE]	Copy a key
DBSIZE	Return number of keys in selected database
DEBUG OBJECT key	Get debugging information about key
DEBUG SEGFAULT	Crash the server
DECR key	Decrement integer value of key by one
DECRBY key decrement	Decrement integer value of key by given number
DEL key [key ...]	Delete a key
DISCARD	Discard all commands issued after MULTI
DUMP key	Return serialized version of value stored at specified key
ECHO message	Echo given string
EVAL script numkeys [key [key ...]] [arg [arg ...]]	Execute Lua script server-side
EVALSHA sha1 numkeys [key [key ...]] [arg [arg ...]]	Execute Lua script server-side
EXEC	Execute MULTI
EXISTS key [key ...]	Check if key exists
EXPIRE key seconds [NX|XX|GT|LT]	Set key time-to-live in seconds
EXPIREAT key timestamp [NX|XX|GT|LT]	Set key expiration time as UNIX timestamp
EXPIRETIME key	Get key expiration Unix timestamp
FAILOVER [TO host port [FORCE]] [ABORT] [TIMEOUT milliseconds]	Start coordinated failover between this server and one of its replicas
FLUSHALL [ASYNC|SYNC]	Delete all keys from all databases
FLUSHDB [ASYNC|SYNC]	Delete all keys from current database
GEOADD key [NX|XX] [CH] longitude latitude member [longitude latitude member ...]	Add one or more geospatial items to geospatial index represented by sorted set
GEOHASH key member [member ...]	Return members of geospatial index as standard geohash strings
GEOPOS key member [member ...]	Return longitude and latitude of geospatial index members
GEODIST key member1 member2 [m|km|ft|mi]	Return distance between two members of geospatial index
GEORADIUS key longitude latitude radius m|km|ft|mi [WITHCOORD] [WITHDIST] [WITHHASH] [COUNT count [ANY]] [ASC|DESC] [STORE key] [STOREDIST key]	Query sorted set representing geospatial index for members within given maximum distance from point
GEORADIUSBYMEMBER key member radius m|km|ft|mi [WITHCOORD] [WITHDIST] [WITHHASH] [COUNT count [ANY]] [ASC|DESC] [STORE key] [STOREDIST key]	Query sorted set representing geospatial index for members within given maximum distance from member
GEOSEARCH key [FROMMEMBER member] [FROMLONLAT longitude latitude] [BYRADIUS radius m|km|ft|mi] [BYBOX width height m|km|ft|mi] [ASC|DESC] [COUNT count [ANY]] [WITHCOORD] [WITHDIST] [WITHHASH]	Query sorted set representing geospatial index for members within box or circle area
GEOSEARCHSTORE destination source [FROMMEMBER member] [FROMLONLAT longitude latitude] [BYRADIUS radius m|km|ft|mi] [BYBOX width height m|km|ft|mi] [ASC|DESC] [COUNT count [ANY]] [STOREDIST]	Query sorted set representing geospatial index for members within box or circle area and store results in another key
GET key	Get value of a key
GETBIT key offset	Return bit value at offset in string value stored at key
GETDEL key	Get value of key and delete the key
GETRANGE key start end	Get substring of string stored at key
GETSET key value	Set string value of key and return its old value
HDEL key field [field ...]	Delete one or more hash fields
HELLO [protover [AUTH username password] [SETNAME clientname]]	Handshake with Redis
HEXISTS key field	Check if hash field exists
HGET key field	Get value of hash field
HGETALL key	Get all fields and values in hash
HINCRBY key field increment	Increment integer value of hash field by given number
HINCRBYFLOAT key field increment	Increment float value of hash field by given amount
HKEYS key	Get all fields in hash
HLEN key	Get number of fields in hash
HMGET key field [field ...]	Get values of all given hash fields
HMSET key field value [field value ...]	Set multiple hash fields to multiple values
HSET key field value [field value ...]	Set string value of hash field
HSETNX key field value	Set hash field value only if field does not exist
HRANDFIELD key [count [WITHVALUES]]	Get one or more random fields from hash
HSTRLEN key field	Get length of hash field value
HVALS key	Get all values in hash
INCR key	Increment integer value of key by one
INCRBY key increment	Increment integer value of key by given amount
INCRBYFLOAT key increment	Increment float value of key by given amount
INFO [section]	Get information and statistics about server
LOLWUT [VERSION version]	Display some computer art and Redis version
KEYS pattern	Find all keys matching given pattern
LASTSAVE	Get UNIX timestamp of last successful save to disk
LINDEX key index	Get element from list by index
LINSERT key BEFORE|AFTER pivot element	Insert element before or after another element in list
LLEN key	Get length of list
LPOP key [count]	Pop and return first element of list
LPOS key element [RANK rank] [COUNT num-matches] [MAXLEN len]	Return index of matching element in list
LPUSH key element [element ...]	Push one or more elements onto list
LPUSHX key element [element ...]	Push element onto list only if list exists
LRANGE key start stop	Get range of elements from list
LREM key count element	Remove elements from list
LSET key index element	Set element in list by index
LTRIM key start stop	Trim list to specified range
MEMORY DOCTOR	Output memory problem report
MEMORY HELP	Display helpful text about different subcommands
MEMORY MALLOC-STATS	Display allocator internal statistics
MEMORY PURGE	Request allocator to release memory
MEMORY STATS	Display memory usage details
MEMORY USAGE key [SAMPLES count]	Estimate memory usage of a key
MGET key [key ...]	Get values of all given keys
MIGRATE host port key| destination-db timeout [COPY] [REPLACE] [AUTH password] [AUTH2 username password] [KEYS key [key ...]]	Atomically transfer key from one Redis instance to another
MONITOR	Monitor all requests received by server in real-time
MOVE key db	Move key to another database
MSET key value [key value ...]	Set multiple keys to multiple values
MSETNX key value [key value ...]	Set multiple keys to multiple values only if none exist
MULTI	Mark start of transaction block
OBJECT ENCODING key	Inspect internal encoding of Redis object
OBJECT FREQ key	Get logarithmic access frequency counter of Redis object
OBJECT IDLETIME key	Get time since Redis object was last accessed
OBJECT REFCOUNT key	Get reference count of key value
OBJECT HELP	Display helpful text about different subcommands
PERSIST key	Remove expiration time from key
PEXPIRE key milliseconds [NX|XX|GT|LT]	Set key time-to-live in milliseconds
PEXPIREAT key milliseconds-timestamp [NX|XX|GT|LT]	Set key expiration time as UNIX timestamp specified in milliseconds
PEXPIRETIME key	Get key expiration Unix timestamp in milliseconds
PFADD key [element [element ...]]	Add specified elements to specified HyperLogLog
PFCOUNT key [key ...]	Return approximate cardinality of set(s) observed by HyperLogLog at key(s)
PFMERGE destkey sourcekey [sourcekey ...]	Merge N different HyperLogLogs into one
PING [message]	Ping server
PSETEX key milliseconds value	Set key value and expiration time in milliseconds
PSUBSCRIBE pattern [pattern ...]	Listen for messages published to channels matching given patterns
PUBSUB CHANNELS [pattern]	List active channels
PUBSUB NUMPAT	Get count of unique pattern subscriptions
PUBSUB NUMSUB [channel [channel ...]]	Get subscriber count for channels
PUBSUB HELP	Display helpful text about different subcommands
PTTL key	Get key time-to-live in milliseconds
PUBLISH channel message	Publish message to channel
PUNSUBSCRIBE [pattern [pattern ...]]	Stop listening for messages published to channels matching given patterns
QUIT	Close connection
RANDOMKEY	Return random key from keyspace
READONLY	Enable read queries for cluster replica node connection
READWRITE	Disable read queries for cluster replica node connection
RENAME key newkey	Rename key
RENAMENX key newkey	Rename key only if new key does not exist
RESET	Reset connection
RESTORE key ttl serialized-value [REPLACE] [ABSTTL] [IDLETIME seconds] [FREQ frequency]	Create key using provided serialized value, previously obtained using DUMP
ROLE	Return role of instance in replication context
RPOP key [count]	Pop and return last element of list
RPOPLPUSH source destination	Pop last element of list, push to another list and return
LMOVE source destination LEFT|RIGHT LEFT|RIGHT	Pop element from list, push to another list and return
RPUSH key element [element ...]	Append one or more elements to list
RPUSHX key element [element ...]	Append element to list only if list exists
SADD key member [member ...]	Add one or more members to set
SAVE	Synchronously save dataset to disk
SCARD key	Get number of members in set
SCRIPT DEBUG YES|SYNC|NO	Set debug mode for executed scripts
SCRIPT EXISTS sha1 [sha1 ...]	Check if script exists in script cache
SCRIPT FLUSH [ASYNC|SYNC]	Remove all scripts from script cache
SCRIPT KILL	Terminate currently executing script
SCRIPT LOAD script	Load specified Lua script into script cache
SDIFF key [key ...]	Subtract multiple sets
SDIFFSTORE destination key [key ...]	Subtract multiple sets and store resulting set in key
SELECT index	Change database selected by current connection
SET key value [EX seconds|PX milliseconds|EXAT timestamp|PXAT milliseconds-timestamp|KEEPTTL] [NX|XX] [GET]	Set string value of key
SETBIT key offset value	Set or clear bit at offset in string value stored at key
SETEX key seconds value	Set key value and expiration time
SETNX key value	Set key value only if key does not exist
SETRANGE key offset value	Overwrite part of string at key starting at specified offset
SHUTDOWN [NOSAVE|SAVE]	Synchronously save dataset to disk and shut down server
SINTER key [key ...]	Return members of set resulting from intersection of all given sets
SINTERCARD numkeys key [key ...] [LIMIT limit]	Intersect multiple sets and return cardinality of result
SINTERSTORE destination key [key ...]	Intersect multiple sets and store resulting set in key
SISMEMBER key member [member ...]	Return whether each member is member of set stored at key
REPLICAOF host port	Make server replica of another instance, or promote it as master
SLOWLOG GET [count]	Get entries from slow log
SLOWLOG LEN	Get length of slow log
SLOWLOG RESET	Clear all entries from slow log
SLOWLOG HELP	Display helpful text about different subcommands
SMEMBERS key	Get all members in set
SMOVE source destination member	Move member from one set to another
SORT key [BY pattern] [LIMIT offset count] [GET pattern [GET pattern ...]] [ASC|DESC] [ALPHA] [STORE destination]	Sort elements in list, set or sorted set
SORT_RO key [BY pattern] [LIMIT offset count] [GET pattern [GET pattern ...]] [ASC|DESC] [ALPHA]	Read-only variant of SORT for sorting elements in list, set or sorted set
SPOP key [count]	Pop and return one or more random members from set
SRANDMEMBER key [count]	Get one or more random members from set
SREM key member [member ...]	Remove one or more members from set
STRLEN key	Get length of value stored at key
SUBSCRIBE channel [channel ...]	Listen for messages published to given channels
SUNION key [key ...]	Return members of set resulting from union of all given sets
SUNIONSTORE destination key [key ...]	Store union of all given sets in key
SWAPDB index1 index2	Swap two Redis databases
SYNC	Internal command for initiating replication stream from master
PSYNC replicationid offset	Internal command for initiating replication stream from master
TIME	Return current server time
TOUCH key [key ...]	Change last access time of keys, return number of specified existing keys
TTL key	Get time-to-live for key in seconds
TYPE key	Determine type stored at key
UNSUBSCRIBE [channel [channel ...]]	Stop listening for messages published to given channels
UNLINK key [key ...]	Asynchronously delete a key in another thread, non-blocking unlike DEL
UNWATCH	Forget about all watched keys
WAIT numreplicas timeout	Wait for synchronous replication of all write commands in current connection context
WATCH key [key ...]	Watch given keys to determine execution of MULTI/EXEC block
ZADD key [NX|XX] [GT|LT] [CH] [INCR] score member [score member ...]	Add one or more members to sorted set, or update scores if they exist
ZCARD key	Get number of members in sorted set
ZCOUNT key min max	Count members in sorted set with scores within given values
ZDIFF numkeys key [key ...] [WITHSCORES]	Subtract multiple sorted sets
ZDIFFSTORE destination numkeys key [key ...]	Subtract multiple sorted sets and store resulting sorted set in new key
ZINCRBY key increment member	Increment score of member in sorted set
ZINTERCARD numkeys key [key ...] [LIMIT limit]	Intersect multiple sorted sets and return cardinality of result
ZINTERSTORE destination numkeys key [key ...] [WEIGHTS weight [weight ...]] [AGGREGATE SUM|MIN|MAX]	Intersect multiple sorted sets and store resulting sorted set in new key
ZLEXCOUNT key min max	Count number of members in sorted set within given lexicographical range
ZPOPMAX key [count]	Pop and return member with highest score from sorted set
ZPOPMIN key [count]	Pop and return member with lowest score from sorted set
ZMPOP numkeys key [key ...] MIN|MAX [COUNT count]	Pop and return members with scores from sorted set
ZRANDMEMBER key [count [WITHSCORES]]	Get one or more random elements from sorted set
ZRANGESTORE dst src min max [BYSCORE|BYLEX] [REV] [LIMIT offset count]	Store range of members from sorted set in another key
ZRANGE key min max [BYSCORE|BYLEX] [REV] [LIMIT offset count] [WITHSCORES]	Return range of members in sorted set
ZRANGEBYLEX key min max [LIMIT offset count]	Return member range in sorted set by lexicographical order
ZREVRANGEBYLEX key max min [LIMIT offset count]	Return member range in sorted set by lexicographical range, ordered high to low by string
ZRANGEBYSCORE key min max [WITHSCORES] [LIMIT offset count]	Return range of members in sorted set by score
ZRANK key member	Determine index of member in sorted set
ZREM key member [member ...]	Remove one or more members from sorted set
ZREMRANGEBYLEX key min max	Remove all members in sorted set within given lexicographical range
ZREMRANGEBYRANK key start stop	Remove all members in sorted set within given index range
ZREMRANGEBYSCORE key min max	Remove all members in sorted set within given score range
ZREVRANGE key start stop [WITHSCORES]	Return range of members in sorted set by index, scores ordered high to low
ZREVRANGEBYSCORE key max min [WITHSCORES] [LIMIT offset count]	Return range of members in sorted set by score, scores ordered high to low
ZREVRANK key member	Determine index of member in sorted set, scores ordered high to low
ZSCORE key member	Get score associated with given member in sorted set
ZMSCORE key member [member ...]	Get scores associated with given members in sorted set
ZUNIONSTORE destination numkeys key [key ...] [WEIGHTS weight [weight ...]] [AGGREGATE SUM|MIN|MAX]	Add multiple sorted sets and store resulting sorted set in new key
SCAN cursor [MATCH pattern] [COUNT count] [TYPE type]	Incrementally iterate keyspace
SSCAN key cursor [MATCH pattern] [COUNT count]	Incrementally iterate Set elements
HSCAN key cursor [MATCH pattern] [COUNT count]	Incrementally iterate hash fields and associated values
ZSCAN key cursor [MATCH pattern] [COUNT count]	Incrementally iterate sorted set elements and associated scores
XINFO CONSUMERS key groupname	List consumers in consumer group
XINFO GROUPS key	List consumer groups for stream
XINFO STREAM key [FULL [COUNT count]]	Get information about stream
XINFO HELP	Display helpful text about different subcommands
XADD key [NOMKSTREAM] [MAXLEN|MINID [=|~] threshold [LIMIT count]] *|ID field value [field value ...]	Append new entry to stream
XTRIM key MAXLEN|MINID [=|~] threshold [LIMIT count]	Trim stream to (approximately if '~' is passed) specified size
XDEL key ID [ID ...]	Remove specified entries from stream, return count of deleted items
XRANGE key start end [COUNT count]	Return range of elements in stream with IDs matching specified ID interval
XREVRANGE key end start [COUNT count]	Return range of elements in stream with IDs matching specified ID interval, in reverse order (from higher to lower IDs) compared to XRANGE
XLEN key	Return number of entries in stream
XREAD [COUNT count] [BLOCK milliseconds] STREAMS key [key ...] ID [ID ...]	Return never seen elements in multiple streams with IDs greater than IDs reported by caller for each stream, can block
XGROUP CREATE key groupname id|$ [MKSTREAM]	Create consumer group
XGROUP CREATECONSUMER key groupname consumername	Create consumer in consumer group
XGROUP DELCONSUMER key groupname consumername	Delete consumer from consumer group
XGROUP DESTROY key groupname	Destroy consumer group
XGROUP SETID key groupname id|$	Set consumer group to arbitrary last delivered ID value
XGROUP HELP	Display helpful text about different subcommands
XREADGROUP GROUP group consumer [COUNT count] [BLOCK milliseconds] [NOACK] STREAMS key [key ...] ID [ID ...]	Return new entries from stream using consumer group, or access history of pending entries for given consumer, can block
XACK key group ID [ID ...]	Mark pending message as correctly processed, effectively removing it from consumer group pending entries list, return count of successfully acknowledged messages
XCLAIM key group consumer min-idle-time ID [ID ...] [IDLE ms] [TIME ms-unix-time] [RETRYCOUNT count] [FORCE] [JUSTID]	In context of stream consumer group, change ownership of pending message so new owner is consumer specified as command argument
XAUTOCLAIM key group consumer min-idle-time start [COUNT count] [JUSTID]	In context of stream consumer group, automatically change ownership of pending message so new owner is consumer specified as command argument
XPENDING key group [[IDLE min-idle-time] start end count [consumer]]	Return information and entries from stream consumer group pending entries list, messages that were read but never acknowledged

AMDC Cluster Usage Guide

AMDC cluster is a high-availability scalable cluster with cluster modes including master-slave mode, sentinel mode, and cluster mode.

Master-Slave Mode

AMDC master-slave mode, also known as Master-Slave Replication, uses one AMDC instance as the master and others as backups. Master and backup data are completely identical. Master supports data writing and reading operations, while slaves support data synchronization and reading from master. When one AMDC fails, other AMDC cache cores can be accessed.

Master-Slave Commands

• Become a slave of a node: slaveof [ip] [port] (replicaof no one will make the slave become master)

Sentinel Mode

Sentinel is an AMDC cache core program that automatically monitors and handles failover between AMDC cache cores. AMDC provides sentinel commands. Sentinels monitor multiple running AMDC instances by sending commands and waiting for AMDC cache core responses.

Sentinel Commands

1. View sentinel status: info
2. Get all master nodes monitored by sentinel: sentinel masters
3. Get status information of master-name master node: sentinel master [master-name]
4. Get status information of all slaves under master-name node: sentinel slaves [master-name]
5. Get IP address by node name in sentinel: sentinel get-master-addr-by-name [master-name]
6. Add node: sentinel monitor [name] [ip] [port] [quorum]
7. Reset amdc name matching specified status: sentinel reset [master-name]
8. Remove node: sentinel remove [master-name]
9. Force node subjective down: sentinel failover [master-name]
10. Get other sentinel node information in sentinel cluster: sentinel sentinels
11. Set access password for specified master node: sentinel set [master-name] auth-pass [password]

Cluster Mode

Cluster mode is AMDC's implementation of elastic scaling for master nodes, mainly used to increase or decrease the memory capacity available to AMDC. It achieves meeting business cache requirements by expanding nodes without adding server memory. Cluster mode also has failover functionality similar to sentinel, with high availability, making it a better choice.

In cluster mode, all nodes can communicate with each other and sense each other's status. The cluster can automatically assign master-slave nodes or manually specify these nodes.

Cluster Commands

1. Assign new hash slots to receiving node: CLUSTER ADDSLOTS slot [slot ...]
2. Command to trigger increment of cluster configuration age from connected node. If node's configuration age is zero or less than cluster's maximum age, age will be incremented: CLUSTER BUMPEPOCH
3. Return count of active failure reports for given node: CLUSTER COUNT-FAILURE-REPORTS node-id
4. Return local key count in specified hash slot: CLUSTER COUNTKEYSINSLOT slot
5. Set hash slot as unbound in receiving node: CLUSTER DELSLOTS slot [slot ...]
6. Force replica to perform manual failover of its master: CLUSTER FAILOVER [FORCE|TAKEOVER]
7. Remove node's own slot information: CLUSTER FLUSHSLOTS
8. Remove node from node table: CLUSTER FORGET node-id
9. Return local key names in specified hash slot: CLUSTER GETKEYSINSLOT slot count
10. Provide information about AMDC Cluster node status: CLUSTER INFO
11. Return hash slot for specified key: CLUSTER KEYSLOT key
12. Force node cluster to handshake with another node: CLUSTER MEET ip port
13. Return node id: CLUSTER MYID
14. Get node's cluster configuration: CLUSTER NODES
15. Set current node as slave of specified master node: CLUSTER REPLICATE [master-node-id]
16. Atomically migrate key to another instance: MIGRATE [host] [port] [key] [destination-db] [timeout] [COPY] [REPLACE] [KEYS key1 key2...]
17. Set slot state (import, export, ownership): CLUSTER SETSLOT [STATE] []
18. Force current node to save current cluster configuration to nodes.conf file immediately: CLUSTER SAVECONFIG
19. Reset node's cluster configuration: CLUSTER RESET [HARD|SOFT]
20. Set node's configuration epoch: CLUSTER SET-CONFIG-EPOCH [epoch]
21. Return all replica node information for specified master node: CLUSTER REPLICAS [node-id]
22. Return mapping information for all slots in cluster, including master and replica nodes for each slot: CLUSTER SLOTS

Cluster Operations Commands

The following commands are auxiliary tools provided by AMDC for daily cluster operations, executed through amdc-cli (non-Cluster protocol commands):

Command	Description
amdc-cli --cluster create ip:port1 ... --cluster-replicas 1	Create cluster (auto meet + assign slots + master-slave)
amdc-cli --cluster check ip:port	Check cluster health status
amdc-cli --cluster info ip:port	View cluster statistics
amdc-cli --cluster fix ip:port	Fix slot coverage issues (experimental)
amdc-cli --cluster reshard ip:port	Online resharding (migrate slots)
amdc-cli --cluster add-node new_ip:new_port existing_host:existing_port	Add new node
amdc-cli --cluster del-node host:port node_id	Delete node

Note: These commands are mainly used to simplify cluster creation, inspection, maintenance and expansion operations, suitable for administrators to manage clusters. --cluster fix and --cluster reshard are advanced features, recommended to fully understand their behavior before use.

SSL Usage

SSL is AMDC's built-in encrypted communication protocol. Enabling SSL enables mutual SSL authentication and encrypted communication with clients. OpenSSL can be used to generate corresponding certificates and key files.

SSL configuration exists in both amdc.yaml and sentinel.yaml with the same usage. For detailed configuration parsing, refer to the cache configuration file section or as follows:

SSL Configuration Items

Based on common parameters for TLS configuration, here is the complete table:

Parameter Name	Description	Usage
tls-port	SSL listening port. If only SSL listening is enabled, set the port under Network to 0	port: 6369
tls-cert-file	Server SSL certificate	tls-cert-file: "./certs/ssl_tls_cert/server.crt"
tls-key-file	Server SSL certificate key	tls-key-file: "./certs/ssl_tls_cert/server.key"
tls-key-file-pass	Server SSL certificate key password	tls-key-file-pass: "password"
tls-ca-cert-file	Certificate authority certificate file, i.e., trusted root certificate	tls-ca-cert-file: "./certs/ssl_tls_cert/ca.crt"
tls-ca-cert-dir	Certificate authority certificate file path. If multiple trusted root certificates exist, use this parameter to configure	tls-ca-cert-dir: ""
tls-client-cert-file	Client SSL certificate, for cluster/master-slave mode	tls-client-cert-file: "./certs/ssl_tls_cert/client.crt"
tls-client-key-file	Client SSL certificate key, for cluster/master-slave mode	tls-client-key-file: "./certs/ssl_tls_cert/client.key"
tls-client-key-file-pass	Client SSL certificate key password	tls-client-key-file-pass: "password"
tls-dh-params-file	Diffie-Hellman parameter file for forward secrecy	tls-dh-params-file: "./certs/ssl_tls_cert/dhparam.pem"
tls-protocols	TLS protocol versions, specify allowed TLS protocol versions	tls-protocols: "TLSv1.2 TLSv1.3"
tls-ciphers	SSL/TLS cipher suites (for TLSv1.2 and below)	tls-ciphers: "ECDHE-RSA-AES256-GCM-SHA384:ECDHE-RSA-AES128-GCM-SHA256"
tls-ciphersuites	TLSv1.3 cipher suites (for TLSv1.3)	tls-ciphersuites: "TLS_CHACHA20_POLY1305_SHA256:TLS_AES_256_GCM_SHA384"
tls-prefer-server-ciphers	Whether to prefer server-defined cipher suites over client-defined	tls-prefer-server-ciphers: "yes"
tls-session-caching	Whether to enable TLS session caching	tls-session-caching: "yes"
tls-session-cache-size	Maximum size of TLS session cache	tls-session-cache-size: 1000
tls-session-cache-timeout	TLS session cache timeout (seconds)	tls-session-cache-timeout: 300
tls-auth-clients	Whether server verifies client certificates. Default requires client to provide certificate and server verifies. Set to "no" if client doesn't need to provide. Set to "optional" for optional client certificate verification	tls-auth-clients: ""
tls-replication	Whether to use TLS for master-slave mode communication. When master parameter is true, slave must also be true	tls-replication: false
tls-cluster	Whether to use TLS for cluster mode communication	tls-cluster: false

Generate Certificates via OPENSSL

Other methods can be used to obtain required certificates as long as they are valid and usable.

For SSL mutual authentication:

• Server needs: CA certificate, server certificate, server private key
• Client needs: CA certificate, client certificate, client private key

Steps:

1. Download and install openssl from https://www.openssl.org/source/index.html

2. openssl.cnf file:

   [ server_cert ]
   keyUsage = digitalSignature, keyEncipherment
   nsCertType = server
   
   [ client_cert ]
   keyUsage = digitalSignature, keyEncipherment
   nsCertType = client
   

3. Create CA certificate:

   • openssl genrsa -out ca.key 4096
   • openssl req -x509 -new -nodes -sha256 -key ca.key -days 3650 -subj "/O=APUSIC /CN=AMDC.com" -out ca.crt

4. Create server private key and certificate

   • openssl genrsa -out amdc.key 2048
   • openssl req -new -sha256 -subj "/O=APUISC /CN=AMD.com" -key amdc.key | openssl x509 -req -CA ca.crt -CAkey ca.key -CAcreateserial -days 365 -extfile openssl.cnf -extensions server_cert -out amdc.crt

5. Create client private key and certificate

   • openssl genrsa -out client.key 2048
   • openssl req -new -sha256 -subj "/O=APUISC /CN=amdc.com" -key client.key | openssl x509 -req -CA ca.crt -CAkey ca.key -CAcreateserial -days 365 -extfile openssl.cnf -extensions server_cert -out client.crt

Server SSL Startup

1. Modify amdc.yaml (key sections):

    TLS/SSL:
      # TLS listening port, use with port 0 to enable TLS only
      tls-port: 0
      # Server TLS certificate file path (PEM format)
      tls-cert-file: "./cert/amdc.crt"
      # Server TLS private key file path (PEM format)
      tls-key-file: "./cert/amdc.key"
      # TLS private key file password (if encrypted)
      tls-key-file-pass: ""
      # Client TLS certificate file path (for client authentication)
      tls-client-cert-file: ""
      # Client TLS private key file path
      tls-client-key-file: ""
      # Client TLS private key file password (if encrypted)
      tls-client-key-file-pass: ""
      # DH parameter file path
      tls-dh-params-file: ""
      # CA certificate file path for verifying clients and nodes
      tls-ca-cert-file: "./cert/cat.crt"
   

2. Start amdc ./amdc-server amdc.yaml

Client SSL Connection

• Using amdc-cli: ./amdc-cli -p 6369 --tls --cert ./client.crt --key ./client.key --cacert ./ca.crt
• Go language connection:

Data Persistence

AMDC provides two persistence methods: RDB and AOF. Related operations are also described in Operation Commands. Here we will expand on the usage and related tools for both persistence methods.

RDB

RDB persistence stores data in AMDC in RDB format to hard disk.

Methods to Generate RDB Files

There are three ways to generate RDB files:

1. Configure save in configuration file to set RDB generation conditions. When triggered, system automatically uses bgsave to generate RDB file
2. Use save command. This command will block all requests, and resume request processing after RDB file is generated
3. Use bgsave command. This command will not block requests, but RDB file generation will be slower than save

Methods to Restore Data from RDB

1. Place one or more RDB files in the directory specified by dir in cache configuration file. Starting amdc will automatically load RDB data.

Repair Corrupted RDB Files

AMDC provides RDB file repair tool

Usage: ./amdc-check-rdb [rdb-filename.rdb]

AOF

AOF persistence writes data in amdc to text files in aof format, stored under appendonlydir folder. amdc will append new requests to the text file.

Methods to Generate AOF Files

1. Change AppendOnly to "yes" in configuration file.

AOF Data Recovery Methods

1. Place one or more AOF files or entire appendonlydir folder in the directory specified by dir in cache configuration file. Starting amdc will automatically load AOF data.

Repair Corrupted AOF Files

AMDC provides AOF file repair tool. However, this tool can only be used if the last appended file is corrupted. This design is because if the base file is already corrupted, most data is lost, and there's no point in repairing.

Usage: ./aof-check-aof --fix [filename.aof|filename.mainfest]

Command Auditing

AMDC provides monitor command to monitor all request information. Combined with shell client, monitoring content can be output to specific files.

Request information recording format: timestamp [database number client address:client port] command

Usage: amdc-cli -h [ip] -p [port] [-a password] monitor >amdc_commands.log

amdc-cli Client Usage Introduction

To facilitate AMDC usage, the shell client provides convenient command-line usage and practical functions such as helping build clusters, manage cluster slots, LRU testing, etc.

Client Parameters

Usage: amdc-cli [parameters] [command [command arguments ...]]

Parameter	Description
-h [hostname]	Cache core IP (default: 127.0.0.1)
-p [port]	Cache core port (default: 6359)
-a [password]	Cache core requirepass password. Can use AMDCCLI_AUTH environment variable to set and input password for better security
--user [username]	Username when logging in with ACL user
--pass [password]	Corresponding password when logging in with ACL user
--askpass	Ignore AMDCCLI_AUTH environment variable, force use of directly entered password
-u [uri]	Cache core URI address (Redis protocol compatible)
-r [repeat]	Repeat specific command [repeat] times
-i [interval]	When using -r, set interval in seconds between repeats
-n [db]	Database number (default 16 databases, numbered 0-15)
-3	Switch to RESP3 protocol
-x	Read input from stdin as last argument for amdc-cli. Example: amdc-cli -x set key < /opt/file
-d [delimiter]	Delimiter between raw format response blocks (e.g.: \n)
-D [delimiter]	Delimiter between multiple raw format responses (e.g.: \n)
-c	Connect to cluster mode
-e	Return error code when command execution fails
--raw	Use raw format output (when tty is not default output device)
--no-raw	Force formatted output even if stdout is not tty
--quoted-input	Force input processing as quoted string
--csv	Output in CSV format
--show-pushes [yn]	Whether to print resp3 push messages, default on
--stat	Dynamically print cache core status: memory/client connections, etc.
--lru-test	Simulate cache workload with 80-20 distribution (key usage)
--replica	Simulate a backup node showing commands received from master
--slave [host:ip]	Specify a node as master of currently connected node
--rdb [filename]	Get RDB file from cache core as local file
--pipe	Transfer raw amdc protocol (Redis compatible) from stdin to cache core
--pipe-timeout [n]	Timeout in pipe mode
--bigkeys	Find keys with large value memory usage
--memkeys	Find key-value pairs with large memory usage
--memkeys-samples	Find key-value pairs with large memory usage, more concise output
--hotkeys	Find frequently used keys, requires lfu cache policy
--scan	List all keys using scan command
--pattern [pat]	Regex pattern matching when using --scan/--bigkeys/--hotkeys, default *
--quoted-pattern [pat]	Similar to --pattern, but string type needs quotes, otherwise considered non-binary safe string
--intrinsic-latency [sec]	System intrinsic latency test, how many seconds test will last
--eval [file]	Send and execute LUA script file
--ldb	Use with --eval, enable amdc lua debugger
--ldb-sync-mode	Same as --ldb, but synchronized with debugger, cache core will be blocked
--cluster [command] [args...] [opts...]	Cluster management commands and arguments. Use --cluster help for detailed commands and arguments
--cluster help	View help information
--verbose	Verbose mode
--no-auth-warning	Don't show warning when using -a to enter password directly
--help	Help information
--version	amdc-cli version information
--tls	Establish a secure TLS connection.
--tls-host	Server name indication for TLS.
--cacert	CA Certificate file to verify with.
--cacert-dir	Directory where trusted CA certificates are stored. If neither cacert nor cacert-dir are specified, the default system-wide trusted root certs configuration will apply.
--insecure	Allow insecure TLS connection by skipping cert validation.
--cert	Client certificate to authenticate with.
--key	Private key file to authenticate with.
--tls-ciphers	Sets the list of preferred ciphers (TLSv1.2 and below) in order of preference from highest to lowest separated by colon ("😊. See the ciphers(1ssl) manpage for more information about the syntax of this string.
--tls-ciphersuites	Sets the list of preferred ciphersuites (TLSv1.3) in order of preference from highest to lowest separated by colon ("😊. See the ciphers(1ssl) manpage for more information about the syntax of this string, and specifically for TLSv1.3 ciphersuites.
--latency	Enter a special mode continuously sampling latency. If you use this mode in an interactive session, it runs forever displaying real-time stats. Otherwise if --raw or --csv is specified, or if you redirect the output to a non TTY, it samples the latency for 1 second (you can use -i to change the interval), then produces a single output and exits.
--latency-history	Like --latency but tracking latency changes over time. Default time interval is 15 sec. Change it using -i.
--latency-dist	Shows latency as a spectrum, requires xterm 256-colors. Default time interval is 1 sec. Change it using -i.

Usage: amdc-cli --cluster [command [command arguments ...]]

Primary Parameter	Secondary Parameter	Description
create host1:port1 ... hostN:portN		Create cluster
	--cluster-replicas [arg]	Number of slave nodes
check host:port		Check cluster
	--cluster-search-multiple-owners	Check if slots are assigned to multiple nodes
info host:port		View cluster status
fix host:port		Fix cluster
	--cluster-search-multiple-owners	Fix slot duplicate assignment issues
reshard host:port		Migrate slots from any node in cluster, reshard slots
	--cluster-from [arg]	Source nodes to migrate slots from, can be multiple nodes separated by comma, passing node id. Can also pass --from all so source nodes are all cluster nodes. If not passed, user will be prompted during migration
	--cluster-to [arg]	Destination node id for slot migration, only one destination node. If not passed, user will be prompted during migration
	--cluster-slots [arg]	Number of slots to migrate. If not passed, user will be prompted during migration
	--cluster-yes	Confirm input during migration
	--cluster-timeout [arg]	Set migrate command timeout
	--cluster-pipeline [arg]	Define number of keys retrieved by cluster getkeysinslot command at once, default 10 if not passed
	--cluster-replace	Whether to directly replace to destination node
rebalance host:port		Balance cluster node slot count from any node
	--cluster-weight [node1=w1...nodeN=wN]	Specify cluster node weights
	--cluster-use-empty-masters	Allow empty master nodes without assigned slots to participate, default not allowed
	--cluster-timeout [arg]	Set migrate command timeout
	--cluster-simulate	Simulate rebalance operation, won't actually execute migration
	--cluster-pipeline [arg]	Define number of keys retrieved by cluster getkeysinslot command at once, default 10
	--cluster-threshold [arg]	Slot migration threshold exceeds threshold, execute rebalance operation
	--cluster-replace	Whether to directly replace to destination node
add-node new_host:new_port existing_host:existing_port		Add node, add new node to specified cluster, default adds master node
	--cluster-slave	New node as slave node, default random master node
	--cluster-master-id [arg]	Specify master node for new node
del-node host:port node_id		Delete specified node, shut down node service after success
call host:port command arg arg .. arg		Execute related commands on all cluster nodes
set-timeout host:port milliseconds		Set cluster-node-timeout
import host:port		Import external amdc data to cluster (non-cluster mode nodes)
	--cluster-from [arg]	Import data from specified instance to cluster
	--cluster-copy	Specify copy when migrate
	--cluster-replace	Specify replace when migrate

Client Usage

AMDC client covers mainstream programming languages such as Java, PHP, Python, C, C++, Node.js, etc. Can connect through amdc-cli, also supports Redis client connecting to amdc for operations, both have same effect.

General Operations

• Connect to amdc-server via URL

  ./amdc-cli -u amdc://user:password@127.0.0.1:6359/db

• Display amdc-cli command help information --help

  Omitted.

• Connect to amdc using amdc-cli client

  ./amdc-cli -h host -p port -a password

  -h specifies IP
  -p specifies port
  -a specifies authentication password

• Output return data to current command line --raw (hide data type) and --no-raw

• Run same command n times consecutively -r, set run interval -i (seconds)

  

  

• Connect to specified db: -n

  

• Execute command output in comma-separated format (CSV format) --csv

  

• Read data from standard input (stdin) as last argument for amdc-cli -x

  

• rdb backup --rdb

  

• Slave mode, treat client as slave node of server --slave

  

  

• Pipeline mode, execute multiple commands at once --pipe

  

Statistical Operations

• Continuous statistics mode, real-time view of amdc key count, memory usage, etc. --stat

  

• Find big keys --bigkeys

  

• Find hot keys --hotkeys

  

Query Operations

• Get all key list --scan

  

• Scan specified keys --pattern (regex)

  

• View minimum, maximum, average access latency of amdc over a period --latency or --latency-history

  

  

Test Operations

• Measure amdc latency within n seconds --intrinsic-latency

  

• LRU simulation, simulate cache eviction policy --lru-test

  

• Simulate showing commands received from master server as replica --replica

  

LUA Operations

• Execute lua script

  

• Start lua debug mode: --ldb or --ldb-sync-mode

  

Cluster Operations

Current multi-tenant version does not support cluster operations

RDB Cluster Data Migration Tool Usage Introduction

Current multi-tenant version does not support cluster operations

Performance Testing Tool Usage Introduction

amdc-benchmark is a dedicated performance testing tool provided to users for testing AMDC, enabling quick understanding of AMDC performance and providing reliable basis for tuning.

AMDC benchmark Parameters

amdc-benchmark

Example: amdc-benchmark [-h <host>] [-p <port>] [-c <clients>] [-n <requests>]

Command parameters

-h <hostname>     Instance host name (default 127.0.0.1)
-p <port>         Instance port (default 6359)
-a <password>     Instance password
-c <clients>      Concurrent connections (default 50)
-n <requests>     Total requests (default 100000)
-d <size>         SET/GET data size in bytes (default 2)
-dbnum <db>       Specify db number (default AMDC has databases 0-15, default 0)
-k <boolean>      1=keep connection 0=reconnect (default 1)
-r <keyspacelen>  SET/GET/INCR use random key, SADD use random value, keyspacelen is max length of random data (max length range 0-12)
-P <numreq>       Merge specified number of requests using pipeline, default 1 (no pipeline)
-e                If server returns error, display in stdout (no more than 1 error per second)
-q                Only display query/sec value
--csv             Output in CSV format
-l                Loop, keep testing
-t <tests>        Specify test commands
-I                Idle mode, open N idle connections and wait
--threads <num>  Enable specified number of threads, default 1

memtier-benchmark parameters

Example: memtier-benchmark [-h <host>] [-p <port>] [-c <clients>] [-n <requests>] [-t <threads>]

Command parameters

-h <hostname>     Instance host name (default 127.0.0.1)
-p <port>         Instance port (default 6359)
-s <socket>       Use Unix domain socket
-a <password>     Authentication password
--tls             Enable TLS encryption
-c <clients>      Number of clients
-t <threads>      Number of threads
--ratio           Read/write ratio
--pipeline <num>  Pipeline command count
--randomdata      Use random data
--hide-histogram  Hide detailed histogram information
--csv <file>      Output csv format file
--test-time <sec> Continuous test duration

Product Configuration Description

Cache Core Configuration

AMDC cache core has three configurable configuration files: cache configuration (amdc.yaml), sentinel configuration (sentinel.yaml), and authorization center configuration (acls.properties).

Cache Configuration File

AMDC cache core configuration file is located at: /installation_root_directory/amdc/amdc.yaml. Some configurations can be modified through console. Below is detailed configuration for AMDC cache core:

Parameter Name	Default Value	Notes
include	empty	Used to include other configuration files, enabling configuration reuse and layered management. CONFIG REWRITE will not modify this option
loadmodule	empty	Load specified module at startup. If loading fails, server will terminate. Can be used multiple times to load multiple modules
bind	127.0.0.1 -::1	Network address to listen on. - prefix means address unavailability won't affect startup. Default only listens on local loopback address. Comment out this line to listen on all interfaces (security risk when exposed to public network)
protected-mode	yes	Protection mode. When no bind specified and no password, only local loopback address and Unix domain socket connections allowed
port	6359	TCP port to listen on. Set to 0 to not listen on TCP socket
tcp-backlog	511	TCP listen backlog size. Actual value may be limited by system somaxconn and tcp_max_syn_backlog
unixsocket	empty	Unix domain socket path. If not specified, won't listen on Unix socket
unixsocketperm	700	Unix domain socket permissions
timeout	0	Client idle timeout (seconds). 0 means disabled
tcp-keepalive	300	TCP keepalive time (seconds). Used to detect dead connections and maintain network connections
license	license.lic	License file path
tls-port	empty	TLS listening port. Use with port 0 to enable TLS only
tls-cert-file	empty	Server TLS certificate file path (PEM format)
tls-key-file	empty	Server TLS private key file path (PEM format)
tls-key-file-pass	empty	TLS private key file password (if encrypted)
tls-client-cert-file	empty	Client TLS certificate file path (for client authentication)
tls-client-key-file	empty	Client TLS private key file path
tls-client-key-file-pass	empty	Client TLS private key file password (if encrypted)
tls-dh-params-file	empty	DH parameter file path. Needed for older OpenSSL (❤️.0), not recommended for newer versions
tls-ca-cert-file	empty	CA certificate file path for verifying clients and nodes
tls-ca-cert-dir	empty	CA certificate directory path for verifying clients and nodes
tls-auth-clients	empty	Client certificate verification policy. no doesn't accept certificates, optional optional but must be valid
tls-replication	no	Whether to enable TLS in replication links
tls-cluster	no	Whether to enable TLS in cluster bus protocol
tls-protocols	"TLSv1.2 TLSv1.3"	Allowed TLS versions, case insensitive
tls-ciphers	DEFAULT:!MEDIUM	Cipher suites allowed for TLSv1.2 and below
tls-ciphersuites	TLS_CHACHA20_POLY1305_SHA256	Cipher suites allowed for TLSv1.3
tls-prefer-server-ciphers	no	Whether to prefer server cipher suite preferences
tls-session-caching	yes	Whether to enable TLS session caching to accelerate reconnections
tls-session-cache-size	20480	Maximum number of TLS session cache entries. 0 means unlimited
tls-session-cache-timeout	300	TLS session cache timeout (seconds)
daemonize	no	Whether to run as daemon. Will generate pid file if yes
supervised	no	Interaction mode with monitoring systems (upstart/systemd). Default no interaction
pidfile	/var/run/amdc_6379.pid	pid file path. Default generated in daemon mode
loglevel	notice	Log level: debug (detailed), verbose (more), notice (moderate, recommended for production), warning (key info only)
logfile	""	Log file path. Empty string means output to stdout
syslog-enabled	no	Whether to enable system log
syslog-ident	amdc	System log identifier
syslog-facility	local0	System log facility. Must be USER or LOCAL0-LOCAL7
crash-log-enabled	yes	Whether to enable crash log
crash-memcheck-enabled	yes	Whether to enable fast memory check in crash log
worker-threads	1	Set number of worker threads. Default 1
databases	16	Number of databases. Default uses DB 0. Can switch via SELECT
always-show-logo	no	Whether to display ASCII art logo in startup log. Default only shown in interactive sessions
set-proc-title	yes	Whether to modify process title to show runtime info
proc-title-template	"{title} {listen-addr} {server-mode}"	Process title template. Supports multiple variables
save	3600 1, 300 100, 60 10000	RDB persistence trigger conditions. Format <seconds> <modifications>. save "" disables RDB
stop-writes-on-bgsave-error	yes	Whether to stop accepting writes when RDB persistence fails
rdbcompression	yes	Whether to use LZF compression for string objects in RDB files
rdbchecksum	yes	Whether to add CRC64 checksum at end of RDB file. Disabling improves performance
sanitize-dump-payload	no	Whether to perform full validation when loading RDB or RESTORE data. Default no due to cluster migration issues
dbfilename	dump.rdb	RDB file name
rdb-del-sync-files	no	Whether to delete replication RDB files when no persistence configured. Only effective when both AOF and RDB disabled
dir	./	Directory for data files (RDB, AOF)
replicaof	empty	Master-slave replication configuration. Format <masterip> <masterport>
masterauth	empty	Master server authentication password
masteruser	empty	Username for replication master server
replica-serve-stale-data	yes	Whether to respond to client requests during replication sync (may return stale data)
replica-read-only	yes	Whether slave server is read-only mode
repl-diskless-sync	no	Whether to enable diskless replication (transfer RDB directly via socket)
repl-diskless-sync-delay	5	Delay time (seconds) to wait for new slave nodes during diskless replication
repl-diskless-load	disabled	How slave loads RDB. disabled (save to disk first), on-empty-db (diskless only when empty), swapdb (keep old data in memory)
repl-ping-replica-period	10	Interval (seconds) for slave to send PING to master
repl-timeout	60	Replication timeout (seconds). Covers SYNC transfer, master-slave heartbeat, etc.
repl-disable-tcp-nodelay	no	Whether to disable TCP_NODELAY for replication links. Yes reduces bandwidth but increases latency
repl-backlog-size	1mb	Replication backlog buffer size. Affects partial resync capability
repl-backlog-ttl	3600	Delay time (seconds) to release backlog buffer after last slave disconnects. 0 means never release
replica-priority	100	Slave priority for Sentinel master selection. 0 means cannot be selected as master
replica-announced	yes	Whether slave is shown in Sentinel reports
min-replicas-to-write	0	Minimum number of slaves required for master to accept writes. 0 disables this feature
min-replicas-max-lag	10	Maximum lag time (seconds) for slaves. Used with min-replicas-to-write
replica-announce-ip	empty	IP address slave announces to master (for NAT scenarios)
replica-announce-port	empty	Port slave announces to master (for NAT scenarios)
tracking-table-max-keys	1000000	Maximum number of keys in client cache tracking table. 0 means unlimited
acllog-max-len	128	Maximum number of ACL log entries
aclfile	empty	External ACL user file path. Cannot be mixed with in-config user configuration
requirepass	empty	Default user password (equivalent to ACL default user password)
acl-pubsub-default	allchannels	Default Pub/Sub channel permissions for new users. allchannels (allow all), resetchannels (deny all)
rename-command	empty	Command renaming (deprecated). Can disable dangerous commands (e.g., rename-command CONFIG "")
maxclients	10000	Maximum concurrent client connections. Limited by system file descriptors
maxmemory	empty	Memory usage limit. When reached, handles according to maxmemory-policy
maxmemory-policy	noeviction	Eviction policy when memory limit reached. E.g., volatile-lru, allkeys-lfu, etc.
maxmemory-samples	5	Number of samples for LRU/LFU/TTL algorithms. Higher is more accurate but uses more CPU
maxmemory-eviction-tenacity	10	Aggressiveness of eviction processing. 0 (lowest latency) to 100 (prioritize eviction)
replica-ignore-maxmemory	yes	Whether slave ignores its own maxmemory configuration (default eviction handled by master)
active-expire-effort	1	Active expired key scan intensity. 1-10, higher is more frequent (may increase CPU and latency)
lazyfree-lazy-eviction	no	Whether to use lazy deletion for memory eviction (background thread frees memory)
lazyfree-lazy-expire	no	Whether to use lazy deletion for key expiration
lazyfree-lazy-server-del	no	Whether to use lazy deletion for internal server key deletion (e.g., RENAME overwrite)
replica-lazy-flush	no	Whether to lazy flush database during slave full sync
lazyfree-lazy-user-del	no	Whether to use lazy deletion for user DEL command (equivalent to UNLINK)
lazyfree-lazy-user-flush	no	Default behavior for user FLUSH command when sync/async not specified
oom-score-adj	no	Whether to control kernel OOM killer process score. yes/relative for relative adjustment, absolute for absolute setting
oom-score-adj-values	0 200 800	OOM score adjustment values for master, slave, background subprocess respectively
disable-thp	yes	Whether to disable kernel Transparent Huge Pages (THP). Avoids fork and CoW performance issues
appendonly	no	Whether to enable AOF persistence
appendfilename	"appendonly.aof"	AOF file name
appendfsync	everysec	AOF fsync strategy: no (depend on system), always (every write), everysec (once per second, default)
no-appendfsync-on-rewrite	no	Whether to disable main process fsync during AOF rewrite. May increase data loss risk but improves performance
auto-aof-rewrite-percentage	100	AOF auto rewrite trigger percentage (current size vs last rewrite growth)
auto-aof-rewrite-min-size	64mb	Minimum file size for AOF auto rewrite
aof-load-truncated	yes	Whether to load truncated AOF file at startup. No means error and exit
aof-use-rdb-preamble	yes	Whether to use RDB prefix during AOF rewrite (improves rewrite and recovery speed)
lua-time-limit	5000	Maximum Lua script execution time (milliseconds). After timeout, only SCRIPT KILL and SHUTDOWN NOSAVE allowed
cluster-enabled	no	Whether to enable cluster mode
cluster-config-file	nodes-6379.conf	Cluster configuration file path. Automatically maintained by node
cluster-node-timeout	15000	Cluster node timeout (milliseconds). Used to determine node failure
cluster-replica-validity-factor	10	Slave failover validity factor. Affects failover conditions
cluster-migration-barrier	1	Slave migration barrier (number of slaves original master must retain)
cluster-allow-replica-migration	yes	Whether to allow slaves to automatically migrate to masters without slaves
cluster-require-full-coverage	yes	Whether cluster requires all hash slots covered to accept requests
cluster-replica-no-failover	no	Whether slaves are prohibited from automatic failover (manual trigger still possible)
cluster-allow-reads-when-down	no	Whether nodes still respond to read requests for their responsible slots when cluster is down
cluster-announce-ip	empty	Cluster node publicly announced IP address (for NAT/Docker scenarios)
cluster-announce-port	empty	Cluster node publicly announced client port
cluster-announce-tls-port	empty	Cluster node publicly announced TLS client port
cluster-announce-bus-port	empty	Cluster node publicly announced bus port (default client port + 10000)
slowlog-log-slower-than	10000	Slow log recording threshold (microseconds). Negative disables. 0 records all commands
slowlog-max-len	128	Maximum number of slow log entries. Oldest entries evicted when exceeded
latency-monitor-threshold	0	Latency monitoring threshold (milliseconds). 0 disables monitoring
notify-keyspace-events	""	Keyspace event notification configuration. Composed of multiple characters (e.g., "Ex" for expiration events)
gopher-enabled	no	Whether to enable Gopher protocol support
hash-max-ziplist-entries	512	Maximum number of entries for hash table using ziplist encoding
hash-max-ziplist-value	64	Maximum value size (bytes) for hash table using ziplist encoding
list-max-ziplist-size	-2	Node size limit for list using ziplist encoding. -1 to -5 corresponds to 4KB to 64KB. Positive numbers are entry count
list-compress-depth	0	List compression depth. 0 disables compression. Positive number means number of uncompressed nodes at head and tail
set-max-intset-entries	512	Maximum number of entries for set using intset encoding (elements are 64-bit signed integers)
zset-max-ziplist-entries	128	Maximum number of entries for sorted set using ziplist encoding
zset-max-ziplist-value	64	Maximum value size (bytes) for sorted set using ziplist encoding
hll-sparse-max-bytes	3000	Maximum bytes for HyperLogLog using sparse representation (includes 16 byte header)
stream-node-max-bytes	4096	Maximum bytes for single node in stream data structure
stream-node-max-entries	100	Maximum number of entries for single node in stream data structure
activerehashing	yes	Whether to enable active rehashing. Periodically frees hash table memory
client-output-buffer-limit normal	0 0 0	Normal client output buffer limit (hard limit, soft limit, soft limit duration)
client-output-buffer-limit replica	256mb 64mb 60	Slave client output buffer limit
client-output-buffer-limit pubsub	32mb 8mb 60	Pub/Sub client output buffer limit
client-query-buffer-limit	1gb	Maximum size of client query buffer
proto-max-bulk-len	512mb	Maximum size of bulk request in protocol (bytes). Must be ≥1mb
hz	10	Background task execution frequency (Hz). 1-500. Higher is more responsive but uses more CPU
dynamic-hz	yes	Whether to enable dynamic Hz. Automatically increases frequency when clients increase
aof-rewrite-incremental-fsync	yes	Whether to fsync every 32MB during AOF rewrite. Reduces latency spikes
rdb-save-incremental-fsync	yes	Whether to fsync every 32MB during RDB save. Reduces latency spikes
lfu-log-factor	10	LFU eviction algorithm counter log factor. Affects counter growth rate
lfu-decay-time	1	LFU counter decay time (minutes). Counter halves after this duration
activedefrag	no	Whether to enable active memory defragmentation (requires Jemalloc support)
active-defrag-ignore-bytes	100mb	Minimum fragmentation waste bytes to trigger active defragmentation
active-defrag-threshold-lower	10	Minimum fragmentation rate (percentage) to trigger active defragmentation
active-defrag-threshold-upper	100	Fragmentation rate (percentage) to trigger maximum defragmentation effort
active-defrag-cycle-min	1	Minimum CPU percentage for defragmentation. Used when lower limit reached
active-defrag-cycle-max	25	Maximum CPU percentage for defragmentation. Used when upper limit reached
active-defrag-max-scan-fields	1000	Maximum fields processed per main dictionary scan (for sets, hashes, etc.)
jemalloc-bg-thread	yes	Whether to enable Jemalloc background thread for memory release
server_cpulist	empty	CPU affinity list for server/IO threads (e.g., "0-7:2" means cores 0, 2, 4, 6)
bio_cpulist	empty	CPU affinity list for background I/O threads
aof_rewrite_cpulist	empty	CPU affinity list for AOF rewrite subprocess
bgsave_cpulist	empty	CPU affinity list for BGSAVE subprocess
ignore-warnings	empty	List of warnings to ignore (space-separated). E.g., "ARM64-COW-BUG"

Sentinel Configuration File

Sentinel configuration file: sentinel.yaml

Parameter Name	Default Value	Notes
bind	127.0.0.1	Network address to listen on. Default only accessible locally. Must explicitly configure (e.g., bind 127.0.0.1 192.168.1.1) or disable protection mode
port	26379	TCP port for sentinel process to listen on
protected-mode	yes	Protection mode. When no bind configured and no password, only local connections allowed. For public network, set to no and use with firewall
unixsocket	empty	Unix domain socket path. If not specified, won't listen on Unix socket
unixsocketperm	700	Unix domain socket permissions
daemonize	no	Whether to run as daemon. Yes means background run and generate pid file
pidfile	/var/run/amdc-sentinel.pid	pid file path. Default generated in daemon mode
logfile	""	Log file path. Empty string means output to stdout. In daemon mode, logs discarded if empty
sentinel announce-ip	empty	For NAT environments. Specifies IP address sentinel publicly announces (replaces auto-detected local IP)
sentinel announce-port	empty	For NAT environments. Specifies port sentinel publicly announces (replaces port config)
dir	/.	Working directory. Sentinel performs persistence operations in this directory
maxclients	10000	Maximum concurrent client connections. Limited by system file descriptors
user	""
sentinel monitor mymaster	127.0.0.1 6379 2	Master node monitoring configuration. Format . Quorum is minimum sentinel count to determine master objectively down
sentinel auth-pass	empty	Authentication password for master and slave nodes. Must set if instance has requirepass configured
sentinel auth-user	empty	Username for sentinel to authenticate with instance (use with auth-pass)
sentinel down-after-milliseconds mymaster	30000	Timeout (milliseconds) for master (or slave, sentinel) to be judged subjectively down (S_DOWN). Default 30 seconds
acllog-max-len	128	Maximum number of ACL log entries. Records commands blocked by ACL and authentication events
aclfile	empty	External ACL user file path. Cannot be mixed with in-config user configuration
requirepass	empty	Sentinel's own authentication password. All sentinels must configure same password to communicate with each other
sentinel sentinel-user	empty	Username for sentinel inter-communication
sentinel sentinel-pass	empty	Password for sentinel inter-communication. Use with sentinel-user
sentinel parallel-syncs mymaster	1	Number of slaves that can be reconfigured as new master's slaves simultaneously during failover. Lower value means less client impact
sentinel failover-timeout mymaster	180000	Failover timeout (milliseconds). Default 3 minutes. Affects retry interval, slave reconfiguration and other scenarios
sentinel notification-script	empty	Sentinel event notification script path (e.g., warning level events). Script receives event type and description as parameters
sentinel client-reconfig-script	empty	Client reconfiguration script path. Triggered after failover to notify clients of master address change
sentinel deny-scripts-reconfig	yes	Whether to prohibit runtime script configuration modification via SENTINEL SET. Default prohibited to avoid security risks
sentinel rename-command	empty	Rename commands (e.g., when master renames CONFIG to GUESSME, sentinel must sync modification)
sentinel resolve-hostnames	no	Whether to enable hostname support. Default only supports IP. Enable requires proper DNS configuration
sentinel announce-hostnames	no	Enable hostname announcement. Use with resolve-hostnames. Default uses IP announcement

Authorization Authentication Center Configuration

When using authentication center, fill in acls.properties. Must provide apusic license authentication center address. Address must be in ip:port format, multiple addresses separated by comma.

Note: This configuration existed in old version's amdc.yaml. Removed in new version but still compatible with old version.

Configuration Item	Default Value	Description
apusic_acls_enable	false	Whether to enable apusic license authentication center. Use true to enable, otherwise false
apusic_acls_authUrls	""	Authentication center address. Must be in ip:port format. Multiple addresses separated by comma
apusic_acls_ns	"public"	Namespace
apusic_acls_tenant	""	Tenant name

Product Tool Usage Instructions

amdc-check-rdb Tool User Manual

Common Issues Handling

amdc-check-rdb is AMDC's RDB file verification and repair tool, used to check the integrity and validity of RDB snapshot files, troubleshoot file corruption issues, and repair some corrupted RDB files to avoid AMDC startup failure or data loss due to RDB file corruption.

Core functions:

• Verify RDB file integrity, detect file format compliance with AMDC specifications
• Repair RDB files corrupted by unexpected interruptions (e.g., server crash, power outage)
• View basic RDB file information (e.g., version, data volume, expiration time, etc.)

Tool Location and Running Prerequisites

Tool Location

After AMDC installation, amdc-check-rdb tool is located in AMDC installation directory folder:

Linux: /amdc/amdc-check-rdb (default path, adjust if custom installation)

Running Prerequisites

• No need to start AMDC service, can run directly from command line
• Must have read permission for RDB file (if insufficient permissions, use chmod to modify, Linux)
• RDB file path must be correct (absolute or relative path. Relative path based on current terminal directory)

Basic Syntax

amdc-check-rdb syntax is simple. Core only needs to specify RDB file path to process. Optional parameters for repair or detailed information:

# Basic syntax (verify RDB file)
amdc-check-rdb [options] /path/to/dump.rdb

# Common shorthand (no options, default only verify file integrity)
amdc-check-rdb dump.rdb

Core Options and Parameters

amdc-check-rdb has few options. Focus on repair and debugging options:

Option	Description	Usage Scenario
--fix	Repair corrupted RDB file, keep recoverable data, discard unrepairable parts	RDB file corrupted, AMDC startup failed
--help (shorthand -h)	Display tool help information, including all option descriptions and basic syntax	Quick lookup when forgetting command syntax

Common Operation Examples

Verify RDB File Integrity

No options needed. Just specify RDB file path. Tool automatically verifies file integrity and format correctness:

# Verify RDB file
amdc-check-rdb /path/to/dump.rdb

Normal output (file intact):

[offset 0] Checking RDB file dump.rdb
[offset 29] AUX FIELD amdc-ver = '2.0.4'
[offset 46] AUX FIELD amdc-bits = '64'
[offset 58] AUX FIELD ctime = '1769518212'
[offset 73] AUX FIELD used-mem = '2442688'
[offset 91] AUX FIELD repl-stream-db = '0'
[offset 141] AUX FIELD repl-id = '069cb616c524d6ec885da6218b669f1d9c970125'
[offset 156] AUX FIELD repl-offset ='0'
[offset 172] AUX FIELD aof-preamble ='0'
[offset 181] Checksum OK
[offset 181] \o/ RDB looks OK! \o/
[info]0 keys read
[info]0 expires
[info] 0 already expired

Abnormal output (file corrupted):

[offset 0] Checking RDB file dump.rdb
[offset 21] AUX FIELD ctime = '1768265496'
[offset 42] AUX FIELD used-mem = '4275241848'
[offset 60] AUX FIELD repl-stream-db = '0'
[offset 11 ] AUX FIELD repl-id = 'f0a6b7b58edlece135f1ba84bcc8add612e8b347'
[offset 135] AUX FIELD repl-offset = '67232446953'
[offset 137] Selecting DB ID 0
--- RDB ERROR DETECTED ---
[offset 73027] Internal error in RDB reading offset 0, function at rdb.c:1936 -> Intset integrity check failed.
[additional info] While doing: read-object-value
[additional info] Reading key 'jszt.894679007767823360.wf.isallownextparticipant.crrccardispatch.audit'
[additional info] Reading type 11 (set-intset)
[info] 159 keys read
[info] 158 expires
[info] 157 already expired
692:C 02 Feb 2026 10:49:23.182 # Terminating server after rdb file reading failure.

Repair Corrupted RDB File

When RDB file is corrupted, use --fix option to repair. Repaired dump file is named dump-<process_id>.rdb:

# Repair corrupted RDB file
amdc-check-rdb --fix /path/to/dump.rdb

Successful repair output:

[offset 0] Checking RDB file --fix
[offset 21] AUX FIELD ctime = '1768265496'
[offset 42] AUX FIELD used-mem = '4275241848'
[offset 60] AUX FIELD repl-stream-db = '0'
[offset 10] AUX FIELD repl-id = 'f0a6b7b58edlece135f1ba84bcc8add612e8b347'
[offset 135] AUX FIELD repl-offset = '67232446953'
[offset 137] Selecting DB ID 0
--- RDB ERROR DETECTED---
[offset 73027] Internal error in RDB reading offset 0, function at rdb.c:1936 -> Intset integrity check failed.
[additional info] While doing: read-object-value
[additional info] Reading key 'jszt.894679007767823360.wf.isallownextparticipant.crrccardispatch.audit'
[additional info] Reading type 11 (set-intset)
[info] 159 keys read
[info] 158 expires
[info] 157 already expired
--- RDB ERROR DETECTED---
[offset 178543] Internal error in RDB reading offset 0, function at rdb.c:1936 -> Intset integrity check failed.
[additional info] While doing: read-object-value
[additional info] Reading key 'jszt.894679007767823360.wf.isallownextparticipant.crrclinefeedback.submit'
[additional info] Reading type 11 (set-intset)
[info] 392 keys read
[info] 390 expires
[info] 388 already expired
--- RDB ERROR DETECTED---
[offset 260382] Internal error in RDB reading offset 0, function at rdb.c:1936 -> Intset integrity check failed.
[additional info] While doing: read-object-value
[additional info] Reading key 'jszt.894679007767823360.wf.isallownextparticipant.ebruleexeccase.save'
[additional info] Reading type 11(set-intset)
[info] 571 keys read
[info] 568 expires
[info] 564 already expired
--- RDB ERROR DETECTED---
[offset 136547728] Invalid object type: 111
[additional info] While doing: read-type
[info] 114982 keys read
[info] 114589 expires
[info] 113699 already expired
[offset 136547728] \o/ RDB partially fixed: dump-790.rdb! \o/
[info] 114982 / 406455 keys fixed

Notes

• Whether to backup original RDB file before repair: --fix option won't directly overwrite original file. Repaired dump file is named dump-<process_id>.rdb. For safety, backup RDB file before operation
• Can only repair partially corrupted files: If RDB file header info is corrupted or checksum is severely wrong, tool may not repair. Need backup files in this case
• Repair may lose some data: For key-value pairs with abnormal format in RDB file, tool will try to skip and discard that data. If unrecoverable data segments detected, will stop scanning remaining file content. Therefore tool may only complete partial file repair
• Tool version must match AMDC version: RDB file formats may differ between versions. Use amdc-check-rdb tool of same version as AMDC service to avoid verification or repair failure

amdc-check-aof Tool User Manual

Tool Introduction

amdc-check-aof is an AOF (Append Only File) file verification and repair tool, used to check AOF file integrity and syntax correctness, repair AOF file corruption caused by unexpected interruptions, ensuring AMDC can normally load AOF file to restore data at startup.

Core functions:

• Verify AOF file syntax correctness and integrity, detect file corruption
• Repair AOF file truncation and syntax errors caused by server crash, power outage, etc.
• Remove invalid commands from AOF file, ensuring AMDC can load normally

Tool Location and Running Prerequisites

Tool Location

Located in same directory as amdc-check-rdb, in AMDC installation folder:

Linux: /amdc/amdc-check-aof

Running Prerequisites

No need to start AMDC service, can run directly from command line

• Must have read and write permissions for AOF file (write needed for repair)
• AOF file path must be correct. If using relative path, ensure terminal current directory matches file path

Basic Syntax

amdc-check-aof syntax is similar to amdc-check-rdb. Core specifies AOF file path. Optional parameters for repair and debugging:

# Basic syntax (verify AOF file)
amdc-check-aof [options] /path/to/appendonlydir/appendonly.aof.manifest

# Common shorthand (no options, default verify file integrity)
amdc-check-aof appendonlydir/appendonlydir/appendonly.aof.manifest

Core Options and Parameters

amdc-check-aof options focus on repair, forced repair and detailed output:

Option	Description	Usage Scenario
--fix	Repair corrupted AOF file, remove invalid commands, truncate error portions. Repaired file overwrites original (backup recommended)	AOF file corrupted, AMDC startup failed
--truncate-to-timestamp	Precisely truncate to specified timestamp, retain all data before that time point	Known service exception time point, need to precisely retain data before specified time
--help (shorthand -h)	Display tool help information, including all options and syntax	Quick lookup when forgetting commands

Common Operation Examples

Verify AOF File Integrity

Directly specify AOF file path. Tool automatically verifies file syntax and integrity:

# Verify AOF file
amdc-check-aof /path/to/appendonlydir/appendonly.aof.manifest

Normal output aof-use-rdb-preamble no (file intact):

Start checking Multi Part AOF
Start to check BASE AOF (RESP format).
AOF analyzed: filename=appendonly.aof.1.base.aof,size=127,ok_up_to=127,ok_up_to_line=22,diff=0
BASE AOF appendonly.aof.1.base.aof is valid
Start to check INCR files.
INCR AOF appendonly.aof.1.incr.aof is empty
All AOF files and manifest are valid

Normal output aof-use-rdb-preamble yes (file intact):

Start checking Multi Part AOF
Start to check BASE AOF (RDB format).
[offset 0] Checking RDB file appendonlydir/appendonly.aof.1.base.rdb
[offset 25] AUX FIELD amdc-ver = '2.0.4'
[offset 38] AUX FIELD amdc-bits = '64'
[offset 50] AUX FIELD ctime = '1770690624'
[offset 65] AUX FIELD used-mem = '836208'
[offset 81] AUX FIELD aof-preamble = '1'
[offset 83] Selecting DB ID 0
[offset 151] Checksum OK
[offset 151] \o/ RDB looks OK! \o/
[info] 2 keys read
[info] 0 expires
[info] 0 already expired
RDB preamble is OK, proceeding with AOF tail...
AOF analyzed: filename=appendonly.aof.1.base.rdb, size=151, ok_up_to=151, ok_up_to_line=1, diff=0
BASE AOF appendonly.aof.1.base.rdb is valid
Start to check INCR files.
INCR AOF appendonly.aof.1.incr.aof is empty
All AOF files and manifest are valid

Abnormal output aof-use-rdb-preamble no (file corrupted):

Start checking Multi Part AOF
Start to check BASE AOF (RESP format).
AOF appendonly.aof.1.base.aof format error
AOF analyzed: filename=appendonly.aof.1.base.aof,size=194,ok_up_to=127,ok_up_to_line=22,diff=67
AOF appendonly.aof.1.base.aof is not valid. Use the --fix option to try fixing it.

Abnormal output aof-use-rdb-preamble yes (file corrupted):

Start checking Multi Part AOF
Start to check BASE AOF (RDB format).
[offset 0] Checking RDB file appendonlydir/appendonly.aof.1.base.rdb
[offset 25] AUX FIELD amdc-ver = '2.0.4'
[offset 38] AUX FIELD amdc-bits = '64'
[offset 50] AUX FIELD ctime = '1770690624'
[offset 65] AUX FIELD used-mem = '836208'
[offset 81] AUX FIELD aof-preamble = '1'
[offset 83] Selecting DB ID 0
[offset 151] Checksum OK
[offset 151] \o/ RDB looks OK! \o/
[info] 2 keys read
[info] 0 expires
[info] 0 already expired
RDB preamble is OK, proceeding with AOF tail...
AOF appendonlydir/appendonly.aof.1.base.rdb format error
AOF analyzed: filename=appendonly.aof.1.base.rdb,size=167,ok_up_to=151,ok_up_to_line=1,diff=16
AOF appendonlydir/appendonly.aof.1.base.rdb is not valid. Use the --fix option to try fixing it.

Repair Corrupted AOF File

When AOF file is corrupted, use --fix option to repair first. Must backup original file before repair:

# 1. Backup original AOF file (Linux)
cp /path/to/appendonlydir /path/to/appendonlydir.bak

# 2. Normal repair
amdc-check-aof --fix /path/to/appendonlydir/appendonly.aof.1.base.aof

Successful repair output aof-use-rdb-preamble no (file corrupted):

Start checking Old-Style AOF
AOF appendonlydir/appendonly.aof.1.base.aof format error
AOF analyzed: filename=appendonlydir/appendonly.aof.1.base.aof,size=146,ok_up_to=127,ok_up_to_line=22,diff=19
This will shrink the AOF appendonlydir/appendonly.aof.1.base.aof from 146 bytes, with 19 bytes, to 127 bytes
Continue? [y/N]: y
Successfully truncated AOF appendonlydir/appendonly.aof.1.base.aof

Successful repair output aof-use-rdb-preamble yes (file corrupted):

Start checking Old-Style AOF
[offset 0] Checking RDB file appendonlydir/appendonly.aof.1.base.rdb
[offset 25] AUX FIELD amdc-ver = '2.0.4'
[offset 38] AUX FIELD amdc-bits = '64'
[offset 50] AUX FIELD ctime = '1770690624'
[offset 65] AUX FIELD used-mem = '836208'
[offset 81] AUX FIELD aof-preamble = '1'
[offset 83] Selecting DB ID 0
[offset 151] Checksum OK
[offset 151] \o/ RDB looks OK! \o/
[info] 2 keys read
[info] 0 expires
[info] 0 already expired
RDB preamble is OK, proceeding with AOF tail...

Notes

• Backup first: Must backup original file before repairing AOF file to avoid complete data loss if repair fails
• Repair principle: Tool repairs files by truncating invalid commands and correcting syntax errors. Unrecognizable commands will be discarded. Check data integrity after repair
• Difference from AOF rewrite: amdc-check-aof is for repairing corrupted files, while AMDC's AOF rewrite (BGREWRITEAOF) is for compressing AOF files. Different functions
• Under appendonlydir directory, can repair specified files via --fix parameter. Supported files include: appendonly.aof.x.base.rdb, appendonly.aof.x.base.aof, appendonly.aof.x.incr.aof

amdc-benchmark Tool User Manual

Tool Introduction

amdc-benchmark is AMDC's built-in performance testing tool, used to simulate multi-client concurrent access to AMDC service, test AMDC performance metrics under different scenarios including requests per second (QPS), response time (latency), concurrent connections, etc., helping operations personnel and developers evaluate AMDC service capacity and optimize configuration parameters to ensure AMDC service runs stably in production environment.

Core functions:

• Test AMDC service throughput (QPS) under different concurrent connection counts and command types
• Evaluate AMDC service response latency (min, max, average latency), troubleshoot performance bottlenecks
• Simulate access scenarios with different data volumes and command combinations, verify AMDC configuration (memory, thread count) reasonableness
• Compare AMDC performance differences under different environments (standalone, cluster) and versions, provide basis for deployment plans

Note: amdc-benchmark is a stress testing tool that will consume AMDC service's CPU, memory and network resources during testing. Use cautiously in production environment. Recommend testing in test environment or during low business periods to avoid affecting online business.

Tool Location and Running Prerequisites

Tool Location

Located in same directory as previous three tools, in AMDC installation folder. No additional installation needed. Deploys with AMDC:

Linux: /amdc/amdc-benchmark

Note: If AMDC installation directory is added to system environment variables, can directly type amdc-benchmark in terminal to run without full path.

Running Prerequisites

• Must start AMDC service: amdc-benchmark needs to connect to AMDC service to execute tests. Ensure amdc-server is running normally
• Network connectivity: Local testing needs 127.0.0.1:6359 (default port) accessible. Remote testing needs AMDC service port (default 6359) open and configured to allow remote connections
• Sufficient permissions: User running amdc-benchmark needs terminal execution permissions. If AMDC service has password, must specify password during testing
• Resource reservation: Reserve sufficient CPU, memory resources before testing. Avoid insufficient test environment resources affecting result accuracy

Basic Syntax

amdc-benchmark syntax core specifies connection parameters and test parameters. Options control test scenarios. Basic syntax:

# Basic syntax (connect AMDC service + execute test)
amdc-benchmark [connection options] [test options]

# Common shorthand (local default connection, no extra options, execute default test)
amdc-benchmark

Notes:

• Connection options: Specify AMDC service host, port, password, etc. Basically same as amdc-cli connection options
• Test options: Control concurrency, test commands, test duration, data volume, etc. Core options for amdc-benchmark
• Default test: If no options specified, tool will test 16 common AMDC commands with 10 concurrent connections, 10000 times each command, outputting performance report

Core Options and Parameters

amdc-benchmark options are divided into "connection options" and "test options". Test options are the focus:

Connection Options

Option	Value Type	Description	Default
-h <host>	String (IP/domain)	Specify AMDC service host address (IP). Can omit for local testing. Must specify for remote testing	127.0.0.1
-p <port>	Integer (port number)	Specify AMDC service port	6359
-s <socket>	String (socket path)	Specify AMDC service password (if service has requirepass configured)	None (default no password)
-a <password>	String	Specify connection timeout (seconds). Avoid connection blocking during testing	None (default no timeout)

Core Test Options

Option	Value Type	Description	Default
-c <clients>	Integer	Specify concurrent client count. Simulates multiple users accessing simultaneously	50
-n <requests>	Integer	Specify total requests per test command. All clients complete together	100000
-d <size>	Integer (bytes)	Specify SET/GET command value data size. Simulates real business value length	3
--dbnum <db>	Integer (database number)	Specify AMDC database to use (SELECT command). Corresponds to AMDC's db0~db15	0
-k <boolean>	Boolean (1/0)	Connection keepalive strategy: 1=keep connection (long connection), 0=reconnect each request (short connection). Simulates real business connection pattern	1
-r <keyspacelen>	Integer	Use random key names for SET/GET/INCR commands, random values for SADD, random members and scores for ZADD. Automatically replaces randint in parameters with 12-digit random number from 0 to keyspacelen-1. Avoids hot keys, fits real scenarios	None
-P <numreq>	Integer	Enable pipeline requests. Each pipeline packages requests to send. Simulates business batch operations (improves throughput)	None
--help	None	Display tool help information including all options and basic syntax	Quick lookup when forgetting command syntax
--version	None	Display amdc-benchmark tool version. Same as AMDC service version	Confirm version compatibility

Advanced Extension Parameters (Multi-threading/Cluster/Special Test Modes)

Option	Value Type	Description	Default
--threads <num>	Integer	Enable multi-threading mode for stress testing tool. Utilizes multi-core CPU to improve stress testing side performance (avoid stress testing side becoming bottleneck)	4
--cluster	None (switch parameter)	Enable cluster mode. Adapts to AMDC Cluster cluster stress testing. Automatically identifies slots and distributes requests	Off
-l	None (switch parameter)	Loop test mode: Run tests infinitely until manually press Ctrl+C to stop. For long-term stability stress testing	Off
-I	None (switch parameter)	Idle mode: Only open N idle connections specified by -c. Don't send any requests. For testing AMDC's idle connection capacity	Off

Common Operation Examples (Core Focus)

Following examples cover most common daily test scenarios. Combined with actual business requirements, adapted for AMDC V2.0.3 and V2.0.4 versions. Can reference directly:

Local Default Test

No options specified. Execute default test: 50 concurrent clients, test 16 default commands, 100000 times each command, output complete performance report:

amdc-benchmark

Default test commands include: SET, GET, INCR, LPUSH, LPOP, SADD, SPOP and other common commands. Suitable for quick understanding of local AMDC basic performance.

Specify Concurrency and Request Count Test

Simulate real concurrent scenarios. Specify concurrent client count and request count per command. Most common testing method:

# Example: 50 concurrent clients, 100000 requests per command, test local AMDC performance
amdc-benchmark -c 50 -n 100000

# Example: 100 concurrent clients, 50000 requests per command, test remote AMDC (with password)
amdc-benchmark -h 192.168.1.100 -p 6359 -a 123456 -c 100 -n 50000

Targeted Core Command Testing

Actual business typically uses only a few AMDC commands. Specify commands via -t option to avoid useless testing and improve efficiency:

# Example: Test SET and GET commands, 50 concurrent, 100000 requests per command
amdc-benchmark -t SET,GET -c 50 -n 100000

# Example: Test hash commands (HSET, HGET), 30 concurrent, value size 1KB
amdc-benchmark -t HSET,HGET -c 30 -n 50000 -d 1024

Time-based Stress Testing

Simulate long-term stable access. Specify test duration. Suitable for testing Redis long-term running stability:

# Example: Loop test (manual stop), test SET command, 60 concurrent, value size 512 bytes
amdc-benchmark -t SET -c 60 -l -d 512

Simplified Output, Quick QPS View

No detailed latency info needed. Only view QPS per command. Suitable for quick performance comparison:

# Example: Default test, simplified output QPS
amdc-benchmark -q

# Example: Test remote AMDC SET, GET commands, simplified output
amdc-benchmark -h 192.168.1.100 -a 123456 -t SET,GET -c 50 -n 100000 -q

Test Result Interpretation (Key Metrics)

After executing test, tool outputs detailed report. Focus on these 4 key metrics. Using SET command as example:

SET: 100000 requests completed in 0.89 seconds
50 parallel clients
3 bytes payload
keep alive:1
host configuration "save": 3600 1 300 100 60 10000
host configuration"appendonly": no
multi-thread:no
Latencyby percentile distribution:
0.000% <= 0.103 ms (cumulative count 123)
50.000% <= 0.215 ms (cumulative count 50123)
90.000% <= 0.327 ms (cumulative count 90123)
99.000% <= 0.439 ms (cumulative count 99012)
99.900% <= 0.551 ms (cumulative count 99901)
99.990% <= 0.663 ms (cumulative count 99990)
99.999% <= 0.775 ms (cumulative count 99999)
100.000% <= 0.887 ms (cumulative count 100000)
Cumulative distribution of latencies:
0.103 ms: 123 requests (0.12%)
0.215 ms: 50123 requests (50.12%)
0.327 ms: 90123 requests (90.12%)
0.439 ms: 99012 requests (99.01%)
0.551 ms: 99901 requests (99.90%)
0.663 ms: 99990 requests (99.99%)
0.775 ms: 99999 requests (99.999%)
0.887 ms: 100000 requests (100.00%)

Summary:
throughput summary: 112359.55 requests per second
latency summary (msec):
| avg   | min   | p50   | p95   | p99   | max   |
|-------|-------|-------|-------|-------|-------|
| 0.220 | 0.080 | 0.215 | 0.383 | 0.439 | 0.887 |

Key metric interpretation:

• throughput summary: QPS (requests per second). Core performance metric. Higher is better (112359.55 QPS in example)
• avg / min / max: Average, minimum, maximum latency (milliseconds). Lower is better
• p50 / p95 / p99: Latency percentiles. Represents 50%/95%/99% of requests have latency not exceeding this value (99% of requests in example have latency ≤ 0.439ms). Percentiles better reflect real performance
• requests completed: Total completed requests. Consistent with -n specified request count. Indicates normal testing

Notes

• Use cautiously in production: amdc-benchmark will consume significant AMDC CPU, memory and network resources. May cause online business lag, timeout. Recommend only in test environment or during low business periods. Stop promptly after testing
• Test environment consistent with production: Try to ensure AMDC configuration (memory, thread count, persistence method), server hardware (CPU, memory) consistent with production during testing. Otherwise test results have no reference value
• Avoid single test scenario: Real business scenarios are complex. Recommend combining concurrency, data volume, command types for multi-scenario testing to comprehensively evaluate performance
• Exclude interference factors: Disable AMDC persistence (appendonly no, save "") during testing to avoid persistence operations affecting test results. Also close other resource-consuming processes on server
• Password security: When using -a option to enter password, password will be displayed in plaintext in terminal history. Recommend clearing terminal history after testing or hiding password via other means
• Reasonably set test parameters: Concurrency should not be too high (exceeding server capacity). Request count/test duration should not be too short (avoid test result fluctuation). Recommend multiple tests and taking average

amdc-conf-conv Tool User Manual

Tool Introduction

amdc-conf-conv is AMDC's dedicated configuration file format conversion tool. Core function is converting AMDC V2.0.2 and below version AMDC configuration files to AMDC V2.0.4 version AMDC configuration files. Solves incompatibility issues between different version AMDC configuration file formats. Enables smooth migration from AMDC V2.0.2 and below to AMDC V2.0.4 without manual configuration modification, improving version migration efficiency and accuracy.

Core functions:

• One-click conversion of AMDC V2.0.2 and below YAML configuration files to AMDC V2.0.4 compatible YAML configuration files. Automatically adapts configuration item formats and naming conventions
• Supports precise conversion of both server and sentinel configuration types. Matches corresponding version functionality configurations
• Preserves all valid configuration items from original configuration file. Ensures converted AMDC V2.0.4 can load and use directly
• Avoids manual conversion configuration syntax errors and missing items. Reduces configuration risks in version migration

Configuration conversion tool execution logic: "First read original structure and basic configuration from --c-yaml template file -> Extract valid configuration items from AMDC V2.0.2 and below configuration file pointed by --go-yaml -> Replace corresponding configuration items in template file with AMDC V2.0.2 and below configuration values -> Finally overwrite and write to file pointed by --c-yaml". Not directly generating new AMDC V2.0.4 configuration file from AMDC V2.0.2 and below configuration. --c-yaml pointed configuration file must not be empty.

Tool Location and Running Prerequisites

Tool Location

Located in same directory as other AMDC supporting tools (like amdc-check-aof, amdc-check-rdb). Deploys with AMDC installation package. No additional installation needed:

Linux: /amdc/amdc-conf-conv

Running Prerequisites

• No need to start AMDC service. Can run directly from command line
• Must have read permission for AMDC V2.0.2 and below configuration file, and read/write permission for target AMDC V2.0.4 configuration file
• Ensure AMDC V2.0.2 and below / AMDC V2.0.4 configuration files (yaml format) paths are correct and files are complete without corruption. Can be read normally

Basic Syntax

amdc-conf-conv syntax rules are fixed. Must specify configuration type and source or target configuration file path. No default execution logic. Must pass complete valid parameters to run. Basic syntax:

# Basic syntax (specify configuration type + source AMDC V2.0.2 and below config + target C version config)
amdc-conf-conv [--server|--sentinel] --go-yaml <amdc_go.yaml> --c-yaml <amdc.yaml>

Core Options and Parameters

amdc-check-aof options focus on repair, forced repair and detailed output:

Option	Description	Usage Scenario
--server	Specify converting AMDC server type configuration file. [--server|--sentinel] is one required parameter. Cannot omit or use simultaneously	Migrating AMDC V2.0.2 and below AMDC server to C version
--sentinel	Specify converting AMDC sentinel type configuration file. [--server|--sentinel] is one required parameter. Cannot omit or use simultaneously	Migrating AMDC V2.0.2 and below AMDC sentinel to C version
--go-yaml	AMDC V2.0.2 and below AMDC configuration file full path (absolute or relative). Specifies conversion source file	All conversion scenarios. Required source file path
--c-yaml	Converted C version AMDC configuration file full path (absolute or relative). Used as replacement template for overwriting	All conversion scenarios. Required target file path
--help	Display tool help information including all options and syntax	Quick lookup when forgetting commands

Common Operation Examples

Following examples cover both server and sentinel core configuration type conversion scenarios. Adapted for Linux environment. Can reference directly. Recommend backing up related configuration files before execution.

Convert AMDC V2.0.2 and below Server Configuration to V2.0.4

Specify --server parameter. Convert AMDC V2.0.2 and below server configuration file conf.yaml to AMDC V2.0.4 server configuration file amdc.yaml:

# Use absolute path (recommended, no path compatibility issues)
amdc-conf-conv --server --go-yaml /amdc/conf.yaml --c-yaml /amdc/amdc.yaml

# Use relative path (ensure terminal current directory is /amdc)
cd /amdc
amdc-conf-conv --server --go-yaml conf.yaml --c-yaml amdc.yaml

Successful conversion:

Skip unsupported configuration item: ReplSlaveIgnoreMaxMemory
Skip unsupported configuration item: PoolSliceSize
Skip unsupported configuration item: PoolSliceNum
Skip unsupported configuration item: PoolNewWorkerNum
Skip unsupported configuration item: AofNoFsyncOnRewrite
Skip unsupported configuration item: GoroutineMaxCpuNum
Skip unsupported configuration item: SlowLogSlowerThan
Skip unsupported configuration item: LuaMaxLocalVarNum
Skip unsupported configuration item: LuaMaxStackSize
Skip unsupported configuration item: ReplServeStaleData
Skip unsupported configuration item: ReplPingSlavePeriod
Skip unsupported configuration item: Ip
Skip unsupported configuration item: Segment
\o/ Config file converted successfully! (conf.yaml -> amdc.yaml)\o/

conf.yaml before conversion:

amdc.yaml before conversion:

amdc.yaml after conversion:

Convert AMDC V2.0.2 and below Sentinel Configuration to V2.0.4

Specify --sentinel parameter. Convert AMDC V2.0.2 and below sentinel configuration file sentinel.yaml to AMDC V2.0.4 sentinel configuration file sentinel.yaml:

# Absolute path example
amdc-conf-conv --sentinel --go-yaml /amdc/sentinelgo.yaml --c-yaml /amdc/sentinel.yaml

# Relative path example (terminal current directory is /amdc)
amdc-conf-conv --sentinel --go-yaml sentinelgo.yaml --c-yaml sentinel.yaml

Successful repair output:

Config file converted successfully! (sentinelgo.yaml -> sentinel.yaml)

sentinel_go.yaml before conversion:

sentinel.yaml before conversion:

sentinel.yaml after conversion:

Notes

• Replacement rule: --c-yaml pointed file is AMDC V2.0.4 configuration replacement template. Tool execution logic: "First read original structure and basic configuration from --c-yaml template file -> Extract valid configuration items from AMDC V2.0.2 and below configuration file pointed by --go-yaml -> Replace corresponding configuration items in template file with AMDC V2.0.2 and below configuration values -> Finally overwrite and write to file pointed by --c-yaml". Not directly generating new AMDC V2.0.4 configuration file from AMDC V2.0.2 and below configuration
• Template file direct overwrite: After tool completes configuration item replacement, will directly overwrite --c-yaml pointed template file without any popup or command line prompt. Must make complete backup of template file before conversion to avoid losing original AMDC V2.0.4 configuration structure or basic items
• Strict parameter rules: Must choose one of --server or --sentinel configuration types. Cannot omit or use simultaneously. --go-yaml (AMDC V2.0.2 and below source config) and --c-yaml (AMDC V2.0.4 replacement template) are required parameters. Must be followed by actual file path. Missing parameters, format errors or invalid paths will trigger tool error and exit
• Conversion direction and type unique: This tool only supports AMDC V2.0.2 and below AMDC config -> AMDC V2.0.4 AMDC config one-way replacement. Does not support AMDC V2.0.4 to AMDC V2.0.2 and below reverse conversion. Not compatible with other version AMDC config replacement. --server or --sentinel must match source file and template file configuration type (server config can only replace server template, sentinel config can only replace sentinel template)
• Must verify validity after replacement: After replacement completes (no error output), must first check if --c-yaml file is properly overwritten and configuration items are replaced, then start corresponding type AMDC V2.0.4 (server or sentinel) to load that file. Confirm service can start normally without configuration related errors to avoid service startup failure due to configuration item matching exceptions

Common Issues Troubleshooting

Tool Cannot Run

Issue 1: Terminal shows "command not found" when entering amdc-check-rdb (Linux)

• Cause: Tool path not added to system environment variable
• Solution: Enter full path to run (e.g., /usr/local/amdc/amdc-check-rdb), or add environment variable (echo "export PATH=/usr/local/amdc/:$PATH" >> /etc/profile, source /etc/profile)

Verification/Repair File Failed

Issue 1: amdc-check-rdb or amdc-check-aof shows "Permission denied"

• Cause: No file read/write permission
• Solution: Use chmod to modify file permissions (chmod 644 dump.rdb), or use sudo to run tool (sudo amdc-check-rdb --fix dump.rdb)

Issue 2: AMDC still cannot start after repair

• Cause: File corruption too severe, or tool version doesn't match AMDC version
• Solution: Use backup file to restore, or upgrade or downgrade tool version to match AMDC service version

amdc-cli Cannot Connect to AMDC Service

Issue 1: Connection shows "Could not connect to AMDC at 127.0.0.1:6359: Connection refused"

• Cause: AMDC service not started
• Solution: Start AMDC service (amdc-server /path/to/amdc.yaml, Linux)

Issue 2: Remote connection shows "Could not connect to AMDC at 192.168.1.100:6359: Connection timed out"

• Cause: Firewall hasn't opened port 6359, or AMDC configuration doesn't allow remote connection
• Solution: Open firewall port (firewall-cmd --add-port=6359/tcp --permanent, Linux), modify AMDC configuration (bind 0.0.0.0, protected-mode no), restart AMDC service

Issue 3: Command execution after connection shows "NOAUTH Authentication required"

• Cause: AMDC service has password set, not entered during connection
• Solution: Enter password using -a option during connection, or enter AUTH password in interactive mode

Summary

This document detailed the usage methods of four AMDC core tools. Key summary:

• amdc-check-rdb: Focused on RDB file verification and repair. Core option --fix. Must backup before repair
• amdc-check-aof: Focused on AOF file verification and repair. Supports normal repair (--fix). Risks require caution
• amdc-benchmark: AMDC's built-in performance testing tool. Used to test AMDC service throughput, response time and other metrics. Core is controlling test scenarios via options to adapt to different performance evaluation needs
• amdc-conf-conv: AMDC's dedicated configuration file format conversion tool. Used to convert AMDC V2.0.2 and below AMDC configuration files to AMDC V2.0.4 AMDC configuration files

All four tools are included with AMDC. No additional installation needed. Must master proficiently in daily operations, especially amdc-cli and amdc-benchmark. Can efficiently complete various AMDC operations, troubleshooting and performance evaluation. During use, must pay attention to data security and testing standards to avoid data loss or service exceptions from misoperation.

Common Issue Resolution

Issue 1: Port Occupation and Resolution

Installing AMDC will occupy some ports. At the same time, client connections depend on internal DNS service. If ports cannot be guaranteed unused, or installation machine environment lacks effective DNS, startup will fail.

Product default occupied ports:

• AMDC cache core default port: 6359
• AMDC sentinel service default port: 26359

If above ports are occupied before startup, startup will fail. Can switch ports or stop programs using those ports.

AMDC Cache Core Port Modification

Modify configuration file: vim amdc.yaml, port field

For more configuration, refer to: Cache Configuration (amdc.yaml)

ACL File Loading

Place ACL file in directory specified by AMDC cache configuration item "ACLFile", then execute ACL load in AMDC cache command line to reload user configuration from ACL file.

Notes:

1. Reloading ACL file will replace existing ACL rules in cache
2. If one or more items in ACL file are invalid, entire file cannot be loaded, existing rules in cache system will continue to be used

Resolving AMDC Master-Slave Failover

1. If slave fails in cache core cluster, system will automatically reconnect When Slave fails and crashes, it will automatically reconnect after recovery. After Master receives Slave connection, it sends complete data file to Slave. If Master receives sync requests from multiple Slaves simultaneously, Master only starts one background process to save data file, then sends to all Slaves, ensuring Slaves function normally.
2. If master fails in cache core cluster, manually restart master service to enable client read/write operations on system When cache core master fails, clients can still read from slave service. Need to manually restart master service. After master restarts and system returns to normal, clients can read/write to master service.

Request Latency Issues

If AMDC requests show obvious latency, can identify specific issues through following methods:

1. slowlog command: Can query commands with longer execution time (duration can be configured via slowlog parameter in amdc.yaml file), better tracking code execution process.
2. –bigkeys: Use amdc-cli client, connect and add –bigkeys parameter to query largest values, determining if large values cause request latency.
3. –memkeys: Same as above. This parameter queries overall key-value memory usage size, determining if large key-value pairs cause request latency.

How to Detect Which Commands Were Executed

Use amdc-cli to connect, then use monitor command to monitor which commands AMDC continuously received. Can also use following command to record to monitor.log in background:

amdc-cli -h [ip] -p [port] [-a password] monitor >monitor.log 2>&1 &

To stop recording monitor.log, use following command:

ps -ef | grep amdc-cli | grep -v grep | awk '{print $2}' | xargs kill -9

Process Exists But Cannot Connect

When clients exceed Linux limits or memory usage exceeds license memory limit, process may exist but connection times out/cannot connect.

1. Clients exceed Linux limits: Some Linux systems default client connection limit to 1024. When business applications create over 1024 client connections, excess will be restricted by Linux and cannot continue connecting.

   Solution: Permanently set user-level file descriptor limits. Add to /etc/security/limits.conf:

   * soft nofile 65535
   * hard nofile 65535
   

   Note: Different Linux systems may have different modification methods. If above doesn't take effect, please search for corresponding system's TCP connection count.

2. Memory usage exceeds license memory limit: License default memory limit is usually 6G (most cases). When business system writes over 6G of data and continues writing, AMDC will frequently trigger memory eviction algorithm, causing AMDC performance degradation. Under sustained high concurrency, some requests may timeout, causing business side errors.

Solution: Scale up to resolve. Two scaling methods: 1. Apply for larger memory license file to directly expand node available memory (not recommended over 16G, too large will reduce query efficiency); 2. Create distributed cluster mode (corresponding to Redis cluster mode).

dir Path Issues

The path configured by "dir" configuration item in configuration file will become the current execution file path after program reads configuration file, not the amdc-server executable file path.

Please note: Method 1: Use absolute paths for certificate paths, log file paths, etc., for stability and reliability. Method 2: If still need to use relative paths, check paths in configuration. Modify certificate, log, etc. paths based on path configured by "dir" configuration item.

Password Setting Issues

When using AMDC master-slave mode, we know requirepass needs to be configured to add password for nodes.

Note: Also need to fill in master node's requirepass for slave node's masterauth configuration item so slave and master nodes can interact normally. Otherwise, if master has requirepass set but slave doesn't have masterauth set, or password is wrong, replication connection will be immediately rejected by master. Logs will show (error) NOAUTH Authentication required. or (error) ERR invalid password errors.

-- Recommendation: Whether cluster service or master-slave service, whether master node or slave node, configure requirepass and masterauth completely to prevent authentication failure issues after master-slave switching.

Similarly, sentinel monitoring also needs to configure node passwords.

Sentinel needs to continuously monitor all master-slave nodes and execute automatic failover when master fails. Therefore, sentinel must also be able to "enter" these password-protected nodes. This is achieved through sentinel auth-pass directive in sentinel's own configuration file: sentinel auth-pass

Log Configuration Issues

System can perform log configuration management on machine for amdc's server.log. Combine with own business to perform simple processing on logs written to server.log by default. Makes logs more personalized and easy to use.

Master-Slave Buffer Configuration Issues

repl-backlog-size is master-slave buffer size. Default configuration is 1mb. Cannot provide buffer function for master-slave services with large data interaction. Recommend increasing to 128mb.