Are you sure the rubies are identical machine-to-machine? I hear that the Ubuntu binaries are pretty terrible and that compiling ruby is the best bet (preferably via RVM).
This might have something to do with garbage collection - many hard-to-diagnose performance issues do. Are you dealing with a ton of Ruby objects whereby the garbage collector might be running mid-action and wreaking havoc? Still though, that is an enormous difference. Are you able to share the code for that particular action? On Thu, Apr 28, 2011 at 4:11 PM, Guyren Howe <[email protected]> wrote: > This has me bamboozled. Check it. This is a typical report running on an > Ubuntu server on slicehost: > > Completed 200 OK in 274704ms (Views: 6039.7ms | ActiveRecord: 931.5ms) > > Here is the same report running on my iMac: > > Completed 200 OK in 4543ms (Views: 2014.2ms | ActiveRecord: 406.8ms) > > And I mean it’s the *same* code. Fresh git checkout. Same version of Rails. > Same version of Postgres and pg gem. Same data. > > Sure, the SliceHost server is slower (it’s a 512MB slice vs a 12GB 3.2GHz > iMac), but it’s not *that* much slower. And it’s not memory bound, either > (checked with top while it was running). The cpu is pegged at 100% the > entire time. > > The log is not terribly helpful. The slowest SQL operation is about half a > second. And it’s not like it’s amazingly long, either, so it’s not stuck in > some loop. And again: same code as runs in 4 seconds on my iMac. > > -- > SD Ruby mailing list > [email protected] > http://groups.google.com/group/sdruby > -- SD Ruby mailing list [email protected] http://groups.google.com/group/sdruby
