Redis - User requested shutdown... - Enhance Control Panel

Redis - User requested shutdown...

MarkD

We are investigating why a WordPress website was offline with the error "Error establishing a Redis connection" and looking at the redis.log file we can see the following entries just before it started:

3:M 11 Mar 2025 11:19:47.975 * Background saving terminated with success
3:M 11 Mar 2025 11:24:48.085 * 100 changes in 300 seconds. Saving...
3:M 11 Mar 2025 11:24:48.097 * Background saving started by pid 6048
6048:C 11 Mar 2025 11:24:52.987 * DB saved on disk
6048:C 11 Mar 2025 11:24:52.996 * Fork CoW for RDB: current 2 MB, peak 16 MB, average 5 MB
3:M 11 Mar 2025 11:24:53.027 * Background saving terminated with success
3:M 11 Mar 2025 11:29:54.076 * 100 changes in 300 seconds. Saving...
3:M 11 Mar 2025 11:29:54.085 * Background saving started by pid 6050
6050:C 11 Mar 2025 11:29:58.963 * DB saved on disk
6050:C 11 Mar 2025 11:29:58.972 * Fork CoW for RDB: current 7 MB, peak 7 MB, average 5 MB
3:M 11 Mar 2025 11:29:58.995 * Background saving terminated with success
3:signal-handler (1741692623) Received SIGTERM scheduling shutdown...
3:M 11 Mar 2025 11:30:23.526 # User requested shutdown...
3:M 11 Mar 2025 11:30:23.526 * Saving the final RDB snapshot before exiting.

It was easy to fix as we simply restarted the PHP container but this is not the first time we have seen redis just stop and wondered if anybody else had seen this?

We haven't changed the config or played around with redis from the stock install so all it has is "bind 127.0.0.1"

Is it even possible for a website user to request redis shut itself down?

cPFence

MarkD

If the user has SSH access, he can run:

redis-cli shutdown

Or maybe Redis got OOM-killed. You can check with:

dmesg | grep -i "oom"

and look at recent logs:

grep -i "redis" /var/log/syslog | tail -50

MarkD

Hi cPFence

The end user would not have accessed the terminal so we know they didn't shut it down themselves. I guess it's possible that they used too much memory but our logs have been truncated so we can't look back that far for any messages.

We have now tweaked the journald settings to retain more logs so we will keep an eye on it.

At least I have something to look for now! Thanks.

bluelinq-computers

MarkD cPFence why not limit redis memory in redis.conf?

MarkD

bluelinq-computers It's a fair suggestion but I don't believe it was a memory issue as our zabbix monitoring platform logs the cpu, memory and process count per website and it wasn't that high just before it stopped. It could have spiked in between reads I guess so yes we should do that.

I know we did play around with the redis config for another website where we switched from using tcp connections to unix sockets and it all ran fine.

bluelinq-computers

MarkD unix sockets all day brother!

ss88us

Does anyone know if turning Redis off then back on in the control panel restarts the PHP container?

I have a similar issue but I am not sure if it's caused by Redis.

Aliysa_Enhance

ss88us Yes, it does restart the PHP container.

ss88us

Aliysa_Enhance Thank you. I don't think my issue is the same!