[03:50] <MFen> SpamapS: ok, at least it's known. when you say "no way".. there is at least one way, which is to reboot. anything i can do short of that? where are pids actually "remembered"?
[03:59] <SpamapS> MFen: yeah, you can exhaust pid space
[03:59] <SpamapS> MFen: they're in RAM
[03:59] <MFen> in init i assume
[04:00] <SpamapS> yeah
[04:00] <MFen> need an initctl cleanup or something
[04:00] <SpamapS> MFen: there's a script attached to the bug which will loop fork/wait until the desired pid appears, then lets upstart kill it
[04:00] <MFen> no reason for upstart to track a pid that doesn't exist
[04:00] <SpamapS> Yeah I agree
[04:00] <MFen> oh heh, that's a clever bit of evil
[04:00] <SpamapS> its by far the highest "Heat" bug on the tracker
[04:01] <SpamapS> bug 406397
[04:01] <MFen> ty
[04:01] <SpamapS> MFen: add yourself as affected. Its basically a vote for what gets worked on by jodh next :)
[04:01] <SpamapS> 51 affected so far
[04:01] <MFen> will do
[04:01] <SpamapS> and its been known for 3+ years
[04:02] <SpamapS> MFen: the only reason it hasn't received love is we keep thinking ther will be a move away from ptrace to something less weird
[04:02] <SpamapS> but then something else comes up..
[04:02] <SpamapS> like logging, or the need to re-exec without forgetting, ironically, all the pids
[04:04] <MFen> just check every 10 loops to see if all the pids still exist, and remove the ones that don't. the solution doesn't need to be fancy
[04:04] <MFen> 52
[04:07] <MFen> wish you could +1 comments on bug trackers. or at least on launchpad
[06:16] <SpamapS> MFen: I agree, or just check all the pids with kill -0 when reload-configuration is asked for