Bug 70967 - Performance regression from cgroup refactoring makes dense boxes nearly unbootable
Summary: Performance regression from cgroup refactoring makes dense boxes nearly unboo...
Status: RESOLVED FIXED
Alias: None
Product: systemd
Classification: Unclassified
Component: general (show other bugs)
Version: unspecified
Hardware: All Linux (All)
: medium critical
Assignee: systemd-bugs
QA Contact: systemd-bugs
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2013-10-28 18:25 UTC by David Strauss
Modified: 2013-10-28 19:41 UTC (History)
0 users

See Also:
i915 platform:
i915 features:

Attachments

Description David Strauss 2013-10-28 18:25:10 UTC 
Having at least few thousand units can add hours to boot time.

We don't have any firm answers yet, but there's a thread here:
http://lists.freedesktop.org/archives/systemd-devel/2013-October/013765.html

It looks like the worst case is a Xen VM because of hypervisor memory management interaction. Like the bottleneck with daemon-reload, key problems seem to be with hashmap_iterate. We've mostly tested with v208, but the changes occurred in v205, the first version to ship after Fedora 19.
Comment 1 David Strauss 2013-10-28 19:41:52 UTC 
Joe (also on the mailing list thread) pointed out that I missed Lennart's reply somehow and the addition to the TODO list. Marking this resolved here.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.