port 6379 daemonize yes Many settings can be changed at runtime using CONFIG GET and CONFIG SET without restarting KeyDB. 3. Network Configuration | Directive | Default | Description | |-----------|---------|-------------| | bind | 127.0.0.1 | Network interfaces to listen on. Use 0.0.0.0 for all. | | port | 6379 | TCP port. Set to 0 to disable TCP. | | tcp-backlog | 511 | Connection backlog size. | | timeout | 0 | Idle connection timeout (seconds). 0 = never. | | tcp-keepalive | 300 | TCP keepalive interval. | | protected-mode | yes | Blocks external access if no bind/password set. | | unixsocket | (none) | Path for Unix socket. | | unixsocketperm | 700 | Permissions for Unix socket. |
| Directive | Default | Description | |-----------|---------|-------------| | lazyfree-lazy-eviction | no | Use async free on eviction. | | lazyfree-lazy-expire | no | Use async free on key expiry. | | lazyfree-lazy-server-del | no | Replace DEL with UNLINK . | | replica-lazy-flush | no | Async flush on replica sync. | | Directive | Default | Description | |-----------|---------|-------------| | slowlog-log-slower-than | 10000 | Microsecond threshold. 0 = all, -1 = off. | | slowlog-max-len | 128 | Max number of slow entries to keep. | 14. Advanced Data Structure Tuning | Directive | Default | Description | |-----------|---------|-------------| | hash-max-ziplist-entries | 512 | Convert hash to ziplist under this many entries. | | hash-max-ziplist-value | 64 | Max size of hash entry value for ziplist (bytes). | | list-max-ziplist-size | -2 | List quicklist ziplist size. | | set-max-intset-entries | 512 | Max integer set entries before conversion. | | zset-max-ziplist-entries | 128 | Max sorted set entries for ziplist. | | zset-max-ziplist-value | 64 | Max value size for ziplist. | | stream-node-max-bytes | 4096 | Max bytes per stream listpack node. | | stream-node-max-entries | 100 | Max entries per stream node. | keydb.cfg
Use active-replica cautiously — it enables optimistic write conflicts. | Directive | Default | Description | |-----------|---------|-------------| | maxclients | 10000 | Max simultaneous client connections. | | client-output-buffer-limit normal | 0 0 0 | Normal clients: hard, soft, seconds. | | client-output-buffer-limit replica | 256mb 64mb 60 | Replica clients. | | client-output-buffer-limit pubsub | 32mb 8mb 60 | Pub/sub clients. | 12. Lazy Freeing (Non-Blocking Deletion) KeyDB supports non-blocking deletes for large keys. port 6379 daemonize yes Many settings can be
| Directive | Default | Description | |-----------|---------|-------------| | server-threads | 2 | Number of threads for handling client requests. | | server-thread-affinity | (none) | CPU core binding for server threads (e.g., 0-3 ). | | io-threads | 1 | Threads for I/O operations (reads/writes). | | io-threads-do-reads | no | Whether I/O threads also handle read commands. | | async-flush | no | Perform flush operations asynchronously. | | async-flush-threads | 4 | Threads for async flush. | | | tcp-backlog | 511 | Connection backlog size
Do not use rename-command to obfuscate commands — use ACLs. Always set a requirepass in production. 8. Memory Management | Directive | Default | Description | |-----------|---------|-------------| | maxmemory | 0 | Max memory in bytes. 0 = unlimited. | | maxmemory-policy | noeviction | Eviction policy: volatile-lru , allkeys-lru , volatile-random , allkeys-random , volatile-ttl , noeviction . | | maxmemory-samples | 5 | Number of keys to sample for LRU/TTL. | | active-defrag | no | Enable online defragmentation. | | active-defrag-threshold-lower | 10 | Fragmentation % to start. | | active-defrag-threshold-upper | 100 | Fragmentation % to force. | | active-defrag-cycle-min | 5 | Minimum CPU % for defrag. | | active-defrag-cycle-max | 75 | Maximum CPU % for defrag. |
For production, bind to specific internal IPs, enable keepalive, and set a reasonable timeout. 4. Threading Model (KeyDB-Specific) This is the core differentiator from Redis.
Increasing server-threads improves throughput on multi-core systems but may increase complexity. Start with cores/2. 5. General Operational Directives | Directive | Default | Description | |-----------|---------|-------------| | daemonize | no | Run as background process. | | pidfile | /var/run/keydb.pid | PID file location. | | loglevel | notice | Levels: debug, verbose, notice, warning. | | logfile | "" | Log file path. Empty = stdout. | | syslog-enabled | no | Log to syslog. | | databases | 16 | Number of logical databases. | | always-show-logo | no | Show KeyDB logo on startup. | | supervised | no | Integration with systemd/upstart. | 6. Persistence Options KeyDB supports both RDB snapshots and AOF (Append-Only File), exactly like Redis. 6.1 RDB (Snapshotting) | Directive | Example | Meaning | |-----------|---------|---------| | save | save 900 1 | Save if ≥1 key changed in 900 sec. | | save | save 300 10 | Save if ≥10 keys changed in 300 sec. | | stop-writes-on-bgsave-error | yes | Stop writes if background save fails. | | rdbcompression | yes | Compress string objects using LZF. | | rdbchecksum | yes | CRC64 checksum at end of RDB. | | dbfilename | dump.rdb | RDB file name. | | dir | ./ | Directory for RDB/AOF files. | 6.2 AOF (Append-Only File) | Directive | Default | Description | |-----------|---------|-------------| | appendonly | no | Enable AOF persistence. | | appendfilename | appendonly.aof | AOF file name. | | appendfsync | everysec | always , everysec , no . | | no-appendfsync-on-rewrite | no | Avoid fsync during rewrite. | | auto-aof-rewrite-percentage | 100 | Trigger rewrite if file grows 100%. | | auto-aof-rewrite-min-size | 64mb | Minimum size for rewrite. | | aof-load-truncated | yes | Load truncated AOF (on crash recovery). | | aof-use-rdb-preamble | yes | Use RDB at start of AOF (hybrid format). |