Brock, I can also see this problem I've times: time until next check -0.001672 and then 0.027262 and then 14.955792
If the time is <= 0 the code is returning True which causes the gobject.idle_add to never finish (if the idle_add calls function returning True it calls it again and again.) So there is recursive call to is_check_required. best Michal Brock Pytlik wrote: > So here's the problem: > > current time Mon 17 Nov 2008 23:51:14 > time until next check -11.827022 > Called do_next_check > time for check: (2008, 11, 17, 23, 51, 14, 0, 322, 0) - 1226965832.64 > > current time Mon 17 Nov 2008 23:51:14 > time until next check -11.827284 > Called do_next_check > time for check: (2008, 11, 17, 23, 51, 14, 0, 322, 0) - 1226965832.64 > > current time Mon 17 Nov 2008 23:51:14 > time until next check -11.827545 > Called do_next_check > time for check: (2008, 11, 17, 23, 51, 14, 0, 322, 0) - 1226965832.64 > > I turned the daly_secs to be 30, so that it would be trying to check for > updates frequently and ran with --debug. Initially, everything seemed to > be working normally, then my CPU usage spiked and I ended up with an > entire scroll buffer of the above form. > > Now that I've got a reproducible test case, I'm going to try a fix, if > that appears to solve the problem, I'll send out a webrev. > > Brock > > > jmr wrote: >> Danek - I'm on mountaineer.sfbay box and checked the >> updatemanagernotifier and schema, both are identical to my own >> machine. I can see the cpu being pegged with prstat. Can you ask Brock >> to: >> $ cat ~/.updatemanager/notify/opensolaris-lastcheck >> $ pstack `pgrep update` >> >> Then do the: >> >> $ pkill updatemanagernotifier >> $ rm ~/.updatemanager/notify/opensolaris-lastcheck >> $ /usr/bin/python2.4 /usr/lib/updatemanagernotifier -d >> >> If this does not chew CPU then come out of it and run it again: >> $ /usr/bin/python2.4 /usr/lib/updatemanagernotifier -d >> >> >> Danek Duvall wrote: >>> On Mon, Nov 17, 2008 at 10:38:49PM +0000, jmr wrote: >>> >>>> I'm not at a build machine, we'll look into it tomorrow. Could you >>>> give Padraig or myself remote access to the box? >>> Brock's in Dan's office debugging the problem, but his machine is >>> mountaineer.sfbay, and is currently exhibiting the problem, if you >>> want to >>> go poking. >>> >>> Danek >> >> _______________________________________________ >> pkg-discuss mailing list >> [email protected] >> http://mail.opensolaris.org/mailman/listinfo/pkg-discuss > _______________________________________________ pkg-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
