RedisRdbLastSaveFailed

Redis tried to write an RDB snapshot in the background and the attempt failed.


Severity	warning
Source	oliver006/redis_exporter 1.50+
Key metric	`redis_rdb_last_bgsave_status` (1 = ok, 0 = last `BGSAVE` failed)

What it means

INFO persistence reports whether the most recent background save succeeded. The alert fires when it has reported failure for several minutes, meaning Redis also has not managed a successful retry.

Two consequences. Your on-disk snapshot is getting older, so a restart loses more data. And with the default stop-writes-on-bgsave-error yes, Redis refuses write commands with MISCONF Redis is configured to save RDB snapshots, but it's currently unable to persist to disk, which is an application outage.

Common causes

Disk full or the data directory is not writable (wrong permissions, read-only volume).
fork() failing with “Cannot allocate memory” because the host lacks memory and vm.overcommit_memory is 0.
Container memory limit too close to dataset size; copy-on-write during the save pushes it over and the child is killed.
dir changed via CONFIG SET to a path that does not exist.

First checks

Read persistence status:

redis-cli -h <host> INFO persistence | grep -E '^(rdb_last_bgsave_status|rdb_bgsave_in_progress|rdb_changes_since_last_save|rdb_last_bgsave_time_sec|rdb_last_cow_size)'
redis-cli -h <host> LASTSAVE

LASTSAVE is the Unix time of the last successful save; date -d @<ts> makes it readable.

Find the reason in the Redis log:

kubectl -n <ns> logs <redis-pod> --since=1h | grep -iE 'background saving|fork|can.t save|error'

Check where it writes and whether there is space:

redis-cli -h <host> CONFIG GET dir
redis-cli -h <host> CONFIG GET dbfilename
df -h <dir>

On the host, check memory overcommit and free memory:
```
sysctl vm.overcommit_memory
free -m
```
See how many writes are unsaved: redis_rdb_changes_since_last_save.

Fixing it

Free or grow the disk, fix directory ownership, or reset dir to a valid path, then trigger redis-cli BGSAVE and confirm the status returns to ok. For fork failures, set vm.overcommit_memory = 1 on the host (the Redis documentation recommends it) and leave memory headroom above the dataset. If writes are blocked and you need them back immediately, CONFIG SET stop-writes-on-bgsave-error no is a temporary bypass, not a fix.

RedisMemoryHigh: tight memory is a common reason fork fails.
RedisDown: a child OOM kill can take the whole container down.
RedisReplicasDisconnected: replica full syncs also depend on successful RDB saves.