INFO

INFO returns the server’s state and statistics as text: sections that start with a # Name line, each followed by field:value lines. The sections and field names follow Redis, so monitoring agents, exporters and dashboards built for Redis read HotKV without changes.

Sections

Without arguments, INFO returns the default sections. Name one or more sections (case-insensitive) to get only those. Unknown names are ignored.

Section	Contents	In default
server	Version, mode, OS, process id, port, uptime, config file.	Yes
clients	Connected, blocked, Pub/Sub and watching clients; largest buffers.	Yes
memory	Memory used, peak, resident size, maxmemory and policy, script memory.	Yes
persistence	Snapshot and append-only file status and statistics.	Yes
stats	Connections, commands, network bytes, hits and misses, expiry and eviction, ACL denials.	Yes
replication	Role, primary link or connected replicas, replication offsets and backlog.	Yes
cpu	CPU time used by the server process.	Yes
modules	Always empty: HotKV has no modules.	Yes
errorstats	Error replies counted by error code.	Yes
cluster	cluster_enabled.	Yes
keyspace	Keys, keys with a TTL, average TTL per database.	Yes
commandstats	Calls, time and failures per command.	No
latencystats	Latency percentiles per command.	No
cluster_info	The same fields as CLUSTER INFO; empty without cluster mode.	No
latency	HotKV only: latency percentiles over all commands together.	No

You can also ask for these groups: default; all, which is every section above except latency; and everything, which is all of them.

Fields you will use most

server: redis_version (the Redis version HotKV is compatible with, 8.4.0, which client libraries check), hotkv_version (the HotKV release), redis_mode (standalone or cluster), os, process_id, run_id, tcp_port, uptime_in_seconds, uptime_in_days, hz, executable, config_file, and one listener<n> line per kind of listener (tcp, tls) with its addresses and port.

clients: connected_clients, maxclients, blocked_clients, pubsub_clients, watching_clients, total_watched_keys, total_blocking_keys, client_recent_max_input_buffer, client_recent_max_output_buffer, tracking_clients (connections with client-side caching tracking on).

memory: used_memory and used_memory_human, used_memory_rss, used_memory_peak, used_memory_peak_perc, used_memory_dataset, used_memory_overhead, total_system_memory, maxmemory, maxmemory_policy, mem_fragmentation_ratio (resident size divided by memory in use), used_memory_lua, used_memory_scripts, number_of_cached_scripts, number_of_functions, number_of_libraries, mem_replication_backlog, mem_clients_normal, mem_clients_slaves (what the clients, and the replicas’ links, hold; see Client memory), mem_not_counted_for_evict (memory maxmemory leaves out: what replicas that lag hold beyond the backlog, and the AOF buffers), mem_aof_buffer, and the mem_cluster_* fields, among them mem_cluster_slot_migration_input_buffer and mem_cluster_slot_migration_input_buffer_peak (the writes a migration target keeps while the copy arrives, see Scaling a cluster).

persistence: rdb_changes_since_last_save, rdb_bgsave_in_progress, rdb_last_save_time, rdb_last_bgsave_status, rdb_saves, aof_enabled, aof_rewrite_in_progress, aof_last_bgrewrite_status, aof_last_write_status, and with the append-only file on, aof_current_size and aof_base_size. loading is always 0.

stats: total_connections_received, total_commands_processed, instantaneous_ops_per_sec, total_net_input_bytes, total_net_output_bytes, instantaneous_input_kbps, instantaneous_output_kbps, rejected_connections, sync_full, sync_partial_ok, sync_partial_err, expired_keys, expired_subkeys, evicted_keys, evicted_clients (clients closed by maxmemory-clients), keyspace_hits, keyspace_misses, pubsub_channels, pubsub_patterns, total_error_replies, total_forks and latest_fork_usec (child processes the server started), the tracking_total_keys, tracking_total_items and tracking_total_prefixes of client-side caching, client_query_buffer_limit_disconnections, client_output_buffer_limit_disconnections, and the acl_access_denied_auth, _cmd, _key and _channel counters.

replication: role (master or slave, as Redis reports it). On a replica: master_host, master_port, master_link_status (up or down), master_last_io_seconds_ago, master_sync_in_progress, slave_repl_offset. On a primary: connected_slaves and one slave<n>:ip=...,port=...,state=...,offset=...,lag=... line per replica. Both: master_replid, master_repl_offset, repl_backlog_active, repl_backlog_size.

cpu: used_cpu_sys, used_cpu_user, used_cpu_sys_children, used_cpu_user_children, in seconds.

keyspace: one line per database that has keys, such as db0:keys=1520,expires=12,avg_ttl=34012,subexpiry=0. avg_ttl is in milliseconds; subexpiry counts keys with field-level expiry.

commandstats: cmdstat_<command>:calls=...,usec=...,usec_per_call=...,rejected_calls=...,failed_calls=....

errorstats: errorstat_<CODE>:count=..., for example errorstat_WRONGTYPE:count=3.

latencystats: latency_percentiles_usec_<command>:p50=...,p99=...,p99.9=..., with the percentiles set by latency-tracking-info-percentiles, while latency-tracking is yes (the default).

latency (HotKV only): latency_p50, latency_p99, latency_p999 in microseconds over every command, and latency_samples.

Fields for Redis features that HotKV does not have are present for compatibility and read 0.

Return value

The text, with \r\n line endings and an empty line between sections: a bulk string in RESP2, a verbatim string (txt) in RESP3. Asking only for unknown sections returns an empty string.

Examples

127.0.0.1:6379> INFO keyspace
# Keyspace
db0:keys=1520,expires=12,avg_ttl=34012,subexpiry=0
db3:keys=8,expires=0,avg_ttl=0,subexpiry=0
127.0.0.1:6379> INFO server clients
# Server
redis_version:8.4.0
hotkv_version:<version>
...
redis_mode:standalone
...
uptime_in_seconds:86400
...

# Clients
connected_clients:12
...

Compatibility

Same section names, order and field names as Redis 7.4, so tools that parse Redis’s INFO work unchanged. Differences: redis_version is the compatibility level and hotkv_version holds the real version; the cluster_info section follows Valkey; the latency section and the everything group are HotKV additions; and fields for features HotKV does not have read 0.

See also

CONFIG RESETSTAT, MEMORY STATS, CLIENT LIST, LATENCY, ROLE, monitoring