Salta al contenuto principale


Lots of Message queues


!Friendica Admins Around 2180 - 43849 for a week or more. Is that very bad? These are my "workers" settings:

Am not sure if I can improve anything and what in this regards, or if I should or not.

in reply to Tio

Are you on the latest pull and is your system working? Then I wouldn't worry.
in reply to Michael Vogel

Yes latest friendica stable and everything works, however some of my messages/posts can take up to 20 minutes to be sent. Pending and pending....but eventually they are sent. I was wondering if I can improve anything.
in reply to Tio

It might help to increase how often the cron job runs, assuming that you aren't already maxing out the server.
in reply to Mark

No the server looks ok. No maxing out.
in reply to Tio

I've noticed that using the worker in daemon mode is far more responsive than the cron job method. Which way do you have it configured?
in reply to Hank G ☑️

Using cron. How did you change to daemon mode?
in reply to Tio

If you’ve pulled develop a couple days ago, this might be relevant to you: https://github.com/friendica/friendica/issues/11778

A couple fixes have been submitted yesterday.
in reply to Hypolite Petovan

Nope I am using the stable :D - am too afraid of the develop :D
in reply to Hypolite Petovan

@Hypolite Petovan I have mixed feelings - partly I'm happy about the current instability of the develop branch; just to have some reminder why the stable branch exists and that develop is basically ALPHA state...

OTOH I'd love to have the usual smooth operation with the develop branch xD
in reply to Tio

I see that they accumulate more and more:
Message queues
2630 - 111040

When is the time to be concerned? :D
in reply to Tio

I would be concerned at that point. Does running in daemon mode instead of cron job mode help burn the queue down faster?
in reply to Hank G ☑️

Now you made me concerned :D - so @Michael Vogel you think the above are worrying numbers? Is there anything I can do to make it faster at dealing with those queues messages?
in reply to Tio

To me the number isn't the problem it is the rate of growth you seem to be reporting and that your system isn't under heavy load trying to work through it. To me that screams either the workers aren't being created fast enough to keep up with it or you are bottlenecked at the DB. The DB bottleneck would show up as either high CPU load or high disk IOPS. Are either of those curves showing lots of thrashing?
in reply to Hank G ☑️

They do not look that bad, I mean the CPU is max at 70% but we have 24 services on this server.

Friendica is not the biggest offender of CPU usage.

in reply to Tio

Which database is Friendica using? The MySQL one? That's using quite a bit of resources. What else is using that database?
in reply to Hank G ☑️

I don't see any Friendica process under mysql:

I also allocate a lot of memory and hardware power to Friendica:

I suppose it uses MySQL / MariaDB.

in reply to Tio

What does the command `mysql -e "SHOW FULL PROCESSLIST";` show?


PS are you on Matrix? This may be easier in the Matrix room since it is more interactive. https://matrix.to/#/#friendi.ca:matrix.org
in reply to Tio

Achtivate these settings in your config file:
        'worker_debug' => true,
        'worker_jpm' => true,
        'worker_jpm_range' => '1',

Then have a look for entries with Load: in it. This tells you about the number of jobs with the different priorities and their processing speed.
in reply to Michael Vogel

Would you mind telling me what config file more exactly? I don't want to ruin anything :D - I suppose it is this https://github.com/friendica/friendica/blob/develop/config/local-sample.config.php - and I have to add those under the "config" section. Correct?
in reply to Tio

@Tio I would suggest you to put friendica in maintenance mode, restart mariadb/mysql and after that optimize the friendica database, using 'mysqloptimize' or 'mysqlcheck -o'.. during the last weeks we saw a lot of 'updatecontact' events and the load skyrocketed.. after optimizing the table, the load and the queue decreased
@Tio

Friendica Admins reshared this.

in reply to Luca Nucifora

Thank you, I will for sure look into this! And btw the link you shared does not work "Access to this profile has been restricted."
Questa voce è stata modificata (1 anno fa)

Friendica Admins reshared this.

in reply to Luca Nucifora

Done that...no change unfortunately. Also moved to daemon from cron. Maybe will wait and see...

Friendica Admins reshared this.

in reply to Tio

@Tio oh, sorry.. I removed it: I wrote some days ago about a similar problem in 'Friendica Admins' group..
if you can, I also suggest you to move from cron to a daemon for the worker queue.. so remember to stop the worker before the 'optimization' procedure
@Tio

Friendica Admins reshared this.

in reply to Luca Nucifora

Ok. I will try some stuff @Hank G ☑️ said he may be able to help me out these days.

Friendica Admins reshared this.