Skip to main content


!Friendica Support
Again I am getting kill messages about the worker

Killed php bin/worker.php

Friendica 'Interrupted Fern' 2024.12 - 1576
php 8.3.10
max parallel running worker: 5

Seems that worker is taking too much RAM and the system (Shared Hoster) is killing the proczess.

reshared this

in reply to Hiker

It might be helpful if you share your full params of current worker-settings
in reply to Hiker

Nothing out of the ordinary here, you could try reducing to less workers but I don't think that will help much. Are you running worker via crons or daemon?

I am running through different hosters (dedicated rack, kvm and also shared hosters) and my hoster does specify things like php process limit, memory limit, script runtime etc...

maybe you can get some more information from your hoster?

This entry was edited (2 months ago)
in reply to Frend

@Frend @Hiker Idea is to put up a simple phpinfo.php with the all-famous 2 lines: <?php phpinfo();. Then copy-paste the details here. Yes, shared hosters have their cheap prices but mostly for limitations, like no federating software with a worker daemon in background. Also other limitations can be CPU usage, RAM and HDD allowed or else you have to upgrade your hosting package to a larger one.

I wrote a "mailer" script, that comes from the old-times of the Internet where you can sign up with such service and collect points to book mails with your points and send them to other members. It has a self-managing part and reduced mail-delivery per click done on the homepage. My then-hoster (shared) asked me to shut it down while it was totally resource-friendly. Just as an example why root servers are often a better choice where you have more freedoms but more responsibilities, like administrating the Linux OS on it.

Friendica Support reshared this.

in reply to Hiker

When its a shared hoster, it might be the case that they have defined a maximum runtime for PHP processes.
in reply to Michael 🇺🇦

The question was about the worker - is it endless as well? "Endless" isn't possible on a shared hoster.
in reply to Hiker

Worker processes (cleanup for example) can easily run for an hour.
in reply to Michael 🇺🇦

So you tell me that to install #Friendica on a shared hoster is a bad idea? Well, root servers are not in my budget.
in reply to Hiker

some very strict webhosters go with 30-60s and then there are webhosters that have like 300-600 seconds. It really depends on your webhoster and the tier/package that you bought. Ask the support or try to find this information in your webhosters package description
in reply to Hiker

I could set it now to 120s - let's see if I will get still these kill messages.
in reply to Hiker

It could also be a problem with too less memory (RAM).
in reply to Hiker

Ok, perhaps the problem is solved with more RAM and longer time in php.ini.
in reply to Hiker

No,I didn't - the kill messages are back again.
in reply to Hiker

All that can be really advised is to check with your hosting provider for shared webspace. Once we have exact rules or limits we can actually try to work around these. Everything else is just a shot in the dark and not really helping much.
in reply to Hiker

The worker kill problem turns out to be a database problem - error log:

Index post-tag is corrupted

How to fix this?

in reply to Hiker

Recreated the table - database error message stopped.
in reply to Hiker

I could set the php run time now to the maximum value of 300s - let's test again.
in reply to Hiker

From both hoster I got the notice that the php execution time is not the problem but the process takes too much RAM.
in reply to Hiker

Do you have messages from the OOM killer in the host's log?
in reply to Rainer "friendica" Sokoll

I have only logs like this from the provider

Feb 05 11:21:20 xyz.server.de kernel: Killed process 22932 (php), UID 153, total-vm:1443352kB, anon-rss:777172kB, file-rss:19936kB, shmem-rss:12472kB

in reply to Hiker

My OOM looks like this:
2025-02-05T17:37:23.376488+01:00 a kernel: [1131349.670942] Out of memory: Killed process 2817607 (ffmpeg) total-vm:50742864kB, anon-rss:17919276kB, file-rss:0kB, shmem-rss:0kB, UID:0 pgtables:61024kB oom_score_adj:0

So, we are in the same boat ;-)