action #39893: warning message about "Failed creating CGroup subtree 'mkdir /sys/fs/cgroup//var: Read-only file system at /usr/lib/perl5/vendor_perl/5.18.2/Mojo/File.pm line 72." - openQA Project - openSUSE Project Management Tool

Actions

Copy link

action #39893

closed

warning message about "Failed creating CGroup subtree 'mkdir /sys/fs/cgroup//var: Read-only file system at /usr/lib/perl5/vendor_perl/5.18.2/Mojo/File.pm line 72."

Added by okurz over 5 years ago. Updated over 5 years ago.

Status:

Resolved

Priority:

Normal

Assignee:

Category:

Regressions/Crashes

Target version:

Start date:

2018-08-17

Due date:

% Done:

Estimated time:

Description

E.g. observed in https://openqa.opensuse.org/tests/734376/file/worker-log.txt

If this "warning" is acceptable it should be phrased in a less severe way or maybe just fixed on o3

Related issues 2 (0 open — 2 closed)

Actions

Copy link

Updated by EDiGiacinto over 5 years ago

Can you tell the systemd version installed in that host? Can you also confirm that the service file of that worker is containing the following line https://github.com/os-autoinst/openQA/blob/master/systemd/openqa-worker%40.service#L19 ?

Originally that message was a (non-fatal), but error.

That basically cames from when openQA tries to setup automatically cgroups (to avoid stale QEMU processes that could be left over from test execution runs). Imho it's needed on worker side to be sure that os-autoinst is not leaving anything behind; so i consider it pretty important feature, as otherwise will bring back things like: https://progress.opensuse.org/issues/6958

Actions

Copy link

Updated by EDiGiacinto over 5 years ago

Related to action #33037: dead children uncaught (again) added

Actions

Copy link

Updated by okurz over 5 years ago

EDiGiacinto wrote:

Can you tell the systemd version installed in that host?

The host is openqaworker1, openSUSE Leap 42.3 with systemd 228-47.1 from openSUSE-Leap-42.3-Update

Can you also confirm that the service file of that worker is containing the following line https://github.com/os-autoinst/openQA/blob/master/systemd/openqa-worker%40.service#L19 ?

Yes, the file /usr/lib/systemd/system/openqa-worker@.service includes Delegate=yes

Originally that message was a (non-fatal), but error.

That basically cames from when openQA tries to setup automatically cgroups (to avoid stale QEMU processes that could be left over from test execution runs). Imho it's needed on worker side to be sure that os-autoinst is not leaving anything behind; so i consider it pretty important feature, as otherwise will bring back things like: https://progress.opensuse.org/issues/6958

I see, make sense to have it warning then. Then however we should investigate why we have it on this host then?

Actions

Copy link

Updated by EDiGiacinto over 5 years ago

okurz wrote:

EDiGiacinto wrote:

Can you tell the systemd version installed in that host?

The host is openqaworker1, openSUSE Leap 42.3 with systemd 228-47.1 from openSUSE-Leap-42.3-Update

Can you also confirm that the service file of that worker is containing the following line https://github.com/os-autoinst/openQA/blob/master/systemd/openqa-worker%40.service#L19 ?

Yes, the file /usr/lib/systemd/system/openqa-worker@.service includes Delegate=yes

Originally that message was a (non-fatal), but error.

That basically cames from when openQA tries to setup automatically cgroups (to avoid stale QEMU processes that could be left over from test execution runs). Imho it's needed on worker side to be sure that os-autoinst is not leaving anything behind; so i consider it pretty important feature, as otherwise will bring back things like: https://progress.opensuse.org/issues/6958

I see, make sense to have it warning then. Then however we should investigate why we have it on this host then?

Thanks for the detailed info, yes this definetly needs investigation, also in new systemd versions, Delegate is not a bool explictly anymore ( so we will have to adapt anyway, as there were changes upstream )

Actions

Copy link