🚨 Redis Error Guide: 'ERR Background save already in progress' — Fix Overlapping BGSAVE Calls

Overview

Redis returns Background save already in progress when you call BGSAVE (or SAVE, or trigger a save via BGREWRITEAOF/DEBUG RELOAD) while an RDB snapshot child process is already running. Redis forks a single child to write the RDB file; it will not fork a second one on top of it.

The literal error clients receive:

(error) ERR Background save already in progress

This is a safety guard, not a fault — Redis is protecting itself from two overlapping saves. It surfaces when a backup script fires BGSAVE while a scheduled save rule already triggered one, or when the previous save is slow because the dataset is large or disk I/O is saturated.

Symptoms

• BGSAVE returns the error while INFO persistence shows a save in flight.
• Correlates with a large dataset, slow disk, or overlapping cron/backup jobs.
• Occasional latency blips as the fork copies-on-write during the save.

redis-cli BGSAVE

(error) ERR Background save already in progress

Common Root Causes

1. A save is genuinely still running

The previous BGSAVE (manual or from a save config rule) has not finished.

redis-cli INFO persistence | grep -E 'rdb_bgsave_in_progress|rdb_last_bgsave_status'

rdb_bgsave_in_progress:1
rdb_last_bgsave_status:ok

2. Backup job races the scheduled save

A cron/backup tool calls BGSAVE at the same time a save 900 1 rule fires.

redis-cli CONFIG GET save

1) "save"
2) "900 1 300 10 60 10000"

3. The save is slow (large dataset / slow disk)

A big dataset or I/O-starved disk makes each fork take long enough for the next request to collide.

redis-cli INFO persistence | grep rdb_last_bgsave_time_sec

4. AOF rewrite is also forking

An in-progress BGREWRITEAOF fork can coincide; Redis serializes these child processes.

How to diagnose

Step 1: Confirm a save is actually in progress

redis-cli INFO persistence | grep -E 'rdb_bgsave_in_progress|aof_rewrite_in_progress'

rdb_bgsave_in_progress:1 means wait — the guard is correct.

Step 2: Check how long saves take

redis-cli INFO persistence | grep -E 'rdb_last_bgsave_time_sec|rdb_last_save_time'

A large rdb_last_bgsave_time_sec explains frequent collisions.

Step 3: Find overlapping triggers

redis-cli CONFIG GET save
sudo journalctl -u redis-server --no-pager | grep -iE 'Background saving|DB saved|Saving' | tail

Compare the log timestamps to your backup schedule.

Step 4: Watch memory/fork cost

redis-cli INFO stats | grep -E 'latest_fork_usec'

A high latest_fork_usec means the fork itself is expensive on this host.

Fixes

Wait for the current save, then retry

The simplest fix — poll rdb_bgsave_in_progress and issue BGSAVE only when it is 0.

while [ "$(redis-cli INFO persistence | tr -d '\r' | awk -F: '/rdb_bgsave_in_progress/{print $2}')" = "1" ]; do
  sleep 1
done
redis-cli BGSAVE

Stop racing the scheduler

Point backups at the RDB file Redis already writes on its save schedule instead of forcing a fresh BGSAVE, or align the cron time so it never overlaps.

Speed up saves

Put the RDB directory on faster disk, ensure enough free RAM for copy-on-write, and consider rdb-save-incremental-fsync yes to smooth I/O.

Use the last-save timestamp to avoid redundant saves

redis-cli LASTSAVE     # only BGSAVE if this is older than your interval

What to watch out for

• BGSAVE forks a child; on a large instance the fork copy-on-write can double memory transiently. Ensure headroom or saves themselves can fail.
• rdb_last_bgsave_status:err (not just “in progress”) points to a failed save — check disk space and permissions, that is a different problem.
• Manual SAVE blocks the whole server; prefer BGSAVE and never call SAVE on a busy production instance.
• Overlapping BGSAVE and BGREWRITEAOF both need a fork; Redis queues them, so tune their schedules apart.

Related

• Redis Error: ‘Background append only file rewriting already in progress’
• Redis Error: ‘Can’t save in background: fork: Cannot allocate memory’
• Redis Error: ‘MISCONF Redis is configured to save RDB snapshots’

Paste the failing command into the free incident assistant, and browse more Redis guides.