Hi,
Has anyone else noticed increased memory usage by the foreman ruby
processes after upgrading? Our foreman servers used to be able to manage 20
odd nodes easily with 1GB memory, after the ugprade they are now maxing out
the memory with foreman:ruby being the main culprit.
Regards,
Andrew.
Yeah, since upgrading to 1.8.1 we've had to double the amount of RAM on our
Foreman server.
Are you able to tell if the number of processes usually running has
increased, or if individual processes are taking up more memory?
Running "passenger-status" is an easy way to check the current number of
processes and the amount of memory they're roughly taking up.
Also check for any stray rake processes, especially when using trends.
–
Dominic Cleal
Red Hat Engineering
Hey,
> Yeah, since upgrading to 1.8.1 we've had to double the amount of RAM on our
> Foreman server.
compared to which version?
> > Hi,
> >
> > Has anyone else noticed increased memory usage by the foreman ruby
> > processes after upgrading? Our foreman servers used to be able to manage
> > 20 odd nodes easily with 1GB memory, after the ugprade they are now
> > maxing out the memory with foreman:ruby being the main culprit.
>
> Are you able to tell if the number of processes usually running has
> increased, or if individual processes are taking up more memory?
>
> Running "passenger-status" is an easy way to check the current number of
> processes and the amount of memory they're roughly taking up.
>
> Also check for any stray rake processes, especially when using trends.
>
Appears to be UI related, perhaps a stray ajax call that leads to a memory
leak. I haven't noticed much now that everything's configured, if someone
logs in and makes about 10+ changes to puppet classes I noticed there was 3
ruby processes (2x foreman, 1x puppet). Right now there's only 1xforeman
and 1x puppet. The additional memory usage didn't die until a reboot or
restart of httpd.
–
Dominic Cleal
Red Hat Engineering–
You received this message because you are subscribed to a topic in the
Google Groups “Foreman users” group.
To unsubscribe from this topic, visit
https://groups.google.com/d/topic/foreman-users/8wPvGXKzdYM/unsubscribe.
To unsubscribe from this group and all its topics, send an email to
foreman-users+unsubscribe@googlegroups.com.
To post to this group, send email to foreman-users@googlegroups.com.
Visit this group at http://groups.google.com/group/foreman-users.
For more options, visit https://groups.google.com/d/optout.
>
> Hey,
>
> > Yeah, since upgrading to 1.8.1 we've had to double the amount of RAM on
> our
> > Foreman server.
>
> compared to which version?
>
I upgraded 3 servers from 1.8.0, 1.8.0 RC3? and 1.7.x
–
Later,
Lukas #lzap Zapletal
> I upgraded 3 servers from 1.8.0, 1.8.0 RC3? and 1.7.x
Can you confirm it's a regression between 1.8 and 1.8.1 ?
Ah okay, so additional processes would make sense as there is more AJAX
in 1.8 versus 1.7.
You may want to try setting and reducing PassengerPoolIdleTime from its
default of 300 seconds
(https://www.phusionpassenger.com/documentation/Users%20guide%20Apache.html)
to ensure that second instance is stopped sooner, but I don't think you
can get away from the need for enough RAM for both processes at once.
The other thing worth mentioning is to ensure your installation is
minimal, as it has a large effect on the memory footprint. If you're
not using some plugins or compute resource support, remove the packages.
–
Dominic Cleal
Red Hat Engineering
>
> > I upgraded 3 servers from 1.8.0, 1.8.0 RC3? and 1.7.x
>
> Can you confirm it's a regression between 1.8 and 1.8.1 ?
>
I wish I could, otherwise I would've opened a bug report instead. I didn't
notice it while on 1.8RC3 or 1.8.0 but I was only on those versions for
about a week.
–
Later,
Lukas #lzap Zapletal
> Can you confirm it's a regression between 1.8 and 1.8.1 ?
Patches we delivered in 1.8.1 are:
http://projects.theforeman.org/rb/issues/backlog/product/foreman?release_id=50
Are you able to identify type of workload that is causing high memory usage? Is it lots of UI requests? Lots of API requests? Lots of incoming reports? Or ENC workload?