-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 > we´ve upgraded puppet from 0.24.8 to 0.25.5 last week, since then we > ´re experiencing load/memory (and then swapping) problems. Our setup > is based on nginx (0.8.20, four worker processes) balancing to five > puppetmaster instances (servertype mongrel, package:rubygem- > mongrel-1.0.1-6.el5), hardware is one quad-core with 12 GB RAM with > approx. 500 clients (puppet runs only once a day), OS is RHEL5.5 64 > Bit. > > this was fine until our upgrade, but now we´re running out of memory > every morning:
this is odd. > nginx active connection peak is ~ 200, memory usage with 0.24.8 was > stable with 2-4 GB. I have 5 instances running using ~400MB of memory. > we´ve not changed anything within our puppet configuration, are there > any topics within > moving 0.24.x to 0.25.x which need to be reconsidered? not that I'm aware of. Actually the upgrade to 0.25.x even reduced memory for me so that I could fire up another mongrel instance ;) What does your setup look like? * are you using custom functions? might they leak? * are you using stored configs? (maybe updating the installed required gems might help) * are all your clients also on 0.25.x or are some still on 0.24.x? I'm aksing because then the xmlrpc api would still be used which might cause a problem * are you using default ruby? Which is 1.8.5 on RHEL 5.5, right? I got also some memory reduction by backporting latest fedora ruby to rhel5.5 (just take the SRPM and rebuild it for rhel5.5) * are you putting out a lot files with puppet? actually 0.25.x should have reduced that footprint heavily, but you never know... maybe you could push them out directly via nginx, see [1] for more. I'm really just guessing and giving you some points where you could look at. I think some of the points shouldn't actually be a problem (like files) but you never know. I ordered them in the way that the most possible in my opinion are on top. You are imho the first person talking about memory problems after switching to 0.25.x but I don't know how special your setup is etc. Can you pinpoint how and when (after 1 client, after 2, after X, after certain clients) your memory grows? I'm pretty sure that if you can elaborate the problem more and maybe get some first points where the problem might be, some puppet-devs might help you more. cheers pete [1] http://www.masterzen.fr/2010/03/21/more-puppet-offloading/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAkwEprYACgkQbwltcAfKi39DzwCeLOmsFOebKRnT0P/00T16pCjA CuQAnjVLvmvMyv2bIh6LkxE7A0OYiXH2 =EAbj -----END PGP SIGNATURE----- -- You received this message because you are subscribed to the Google Groups "Puppet Users" group. To post to this group, send email to puppet-us...@googlegroups.com. To unsubscribe from this group, send email to puppet-users+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/puppet-users?hl=en.
