On 04/08/09 at 23:37 +0000, Stephen Gran wrote: > Package: ruby1.8 > Version: 1.8.7.72-3 > Severity: important > X-Debbugs-cc: [email protected] > > Hi there, > > I've just managed to capture a reasonable strace of a problem that we've > been seeing with puppet on several platforms. This is on alpha, but I'm > fairly sure we've also seen it on armel and possibly hppa as well. > Basically, sleep doesn't sleep. An eventloop with a wakeup timer just > ends up in a tight loop, which means our slower machines can't do > anything except run a single ruby thread that is supposed to be > sleeping. > > I am assuming this is a bug in the thread/select implementation in ruby, > but feel free to reassign if it turns out to be libc/kernel/whatever > related. > > Any bugs in the test code are presumably mine - my ruby is fairly mediocre > at best, and my goal was not to write a perfect event loop but to > reproduce as faithfully as possible the event loop that puppet uses > without actually doing anything useful besides the event loop.
Hi, I haven't had time to look into that bug yet (I'm supposed to be in VAC). However, it would be interesting to see if it is reproducible with ruby1.9 (the lenny version would do). Ruby's threading was completely re-implemented in 1.9, so it is likely not to have the same bugs ;) -- | Lucas Nussbaum | [email protected] http://www.lucas-nussbaum.net/ | | jabber: [email protected] GPG: 1024D/023B3F4F | -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
