On Tue, Oct 25, 2022 at 01:13:44PM +0100, Daniel P. Berrangé wrote: > On Tue, Oct 25, 2022 at 01:46:22PM +0200, Michal Prívozník wrote: > > On 10/25/22 13:43, Pavel Hrdina wrote: > > > This reverts commit e49313b54ed2a149c71f9073659222742ff3ffb0. > > > This reverts commit a0f37232b9c4296ca16955cc625f75eb848ace39. > > > > > > Revert them together to not break build. > > > > > > This fix of the issue is incorrect and breaks usage of other controllers > > > in hybrid mode that systemd creates, specifically usage of devices and > > > cpuacct controllers as they are now assumed to be part of the cgroup v2 > > > topology which is not true. > > > > > > We need to find different solution to the issue. > > > > > > Signed-off-by: Pavel Hrdina <phrdina@xxxxxxxxxx> > > > --- > > > src/util/vircgroup.c | 6 ++---- > > > src/util/vircgroupv2.c | 15 +++++++++++++++ > > > 2 files changed, 17 insertions(+), 4 deletions(-) > > > > > > Reviewed-by: Michal Privoznik <mprivozn@xxxxxxxxxx> > > > > This brings me back to my original proposal - let's forget that hybrid > > mode exist and aim on v1 or v2, solely. > > FWIW, I thought that was already our intention. It has been recommended > to me by systemd maintainers that hybrid mode is best ignored. The issue is that on systemd based distributions where you have v1 only it is technically hybrid setup where systemd uses v2 without any controller enabled for process tracking and that's the root cause of the issue that the original patch wanted to address. We need to add the lxc-container process to the correct cgroup in order to be able to find the cgroup by PID of the lxc-container using systemd API. True v1 mode is enabled only on few distributions without systemd. Pavel
Attachment:
signature.asc
Description: PGP signature
