Re: akonadiserver and kmail are leaking memory
Wolfgang Bauer <[email protected]>
| Newsgroups | gmane.linux.suse.kde |
|---|---|
| Message-ID | <[email protected]> |
Am Sonntag, 31. März 2019, 23:58:03 schrieb stakanov: > In data domenica 31 marzo 2019 22:59:05 CEST, Hans-Peter Jansen ha scritto: > > Am Sonntag, 31. März 2019, 19:38:14 CEST schrieb stakanov: > > > In data domenica 31 marzo 2019 18:05:32 CEST, Hans-Peter Jansen ha > > scritto: > > > > Yes, that reads 41G VSZ and 27G RSS for akondiserver, only in about 5 > > > > hours > > > > wall clock run time. > > > > > > Leap 15.0 > > > Bug 1100080 > > > Try to uninstall baloo indexer. When it crashes then kmail leaks huge > > > amount of memory up to shutdown or unresponsiveness. > > > The bug is known to kde and it is linked to lmdb. > > > While to TW it was cherrypicked and backported (there is a fix) the Leap > > > 15 > > > version did not receive this blessing up to now. > > > with mysql, the system does not complain always, only sometimes rarely > > > konqi pops up after baloo crashed. > > > > Don't know, if that fits the bill. > > > > It's Tumbleweed, and if I'm reading your report correctly, the lmdb issue > > is fixed here. Looks, like upstream found it even severe enough to create > > new release, that is awaiting TW blessing, but a patch with a fix was > > applied Oct 6, 2018. > > > > Next, it's the kernel, that stops akonadiserver, when it has eaten all > > memory and swap. I don't let it swap, when I'm on, of course. > > > > The baloo processes behave, they use 256G VSZ, but only tiny amounts of > > RSS, which is the painful part here. > > > > I need to stop now, and shutdown akonadi again, before the system becomes > > sluggish and this mail wont get out... > > > > Cheers, > > Pete > > You have my compassion. That problem must be fixed. One bug report is here: https://bugs.kde.org/show_bug.cgi?id=395131 (but I think I have seen other similar ones as well) Of course it should be fixed, but one problem is that it certainly doesn't happen in general, so it would need to be determined what exactly triggers it... JFTR, I don't see this on Leap 15.0 or 42.3 either, so it's probably not a general problem with the mariadb versions used as one comment in that bug report suggests (and my Akonadi databases are quite old, one even dates back to KDEPIM 4.7, but that one actually uses an external MySQL/MariaDB database). Kind Regards, Wolfgang -- To unsubscribe, e-mail: [email protected] To contact the owner, e-mail: [email protected]