Page MenuHomePhabricator

[ceph] Test and upgrade to kernel ~15
Closed, ResolvedPublic

Description

There's some performance improvements on this newer kernel that would help with the osds, full changelog:
https://cdn.kernel.org/pub/linux/kernel/v5.x/ChangeLog-5.10.17

Recommendation comes from our ceph contractor, as they run this major version on all their installations and saw a significant improvement.

Current results don't show any noticeable improvement:

Related Objects

StatusSubtypeAssignedTask
Resolveddcaro
Resolveddcaro

Event Timeline

dcaro triaged this task as High priority.

Change 674074 had a related patch set uploaded (by David Caro; owner: David Caro):
[operations/puppet@production] wmcs.ceph.codfw: Upgrade to latest 5.X kernel

https://gerrit.wikimedia.org/r/674074

Change 674074 merged by David Caro:
[operations/puppet@production] wmcs.ceph.codfw: Upgrade to latest 5.X kernel

https://gerrit.wikimedia.org/r/674074

Mentioned in SAL (#wikimedia-cloud) [2021-03-25T10:08:49Z] <dcaro> upgrading kernel on cloudcephmon2003-dev and reboot (T274565)

Mentioned in SAL (#wikimedia-cloud) [2021-03-25T10:18:23Z] <dcaro> upgrading kernel on cloudcephosd2002-dev and reboot (T274565)

Mentioned in SAL (#wikimedia-cloud) [2021-03-25T10:24:10Z] <dcaro> upgrading kernel on cloudcephosd2003-dev and reboot (T274565)

Mentioned in SAL (#wikimedia-cloud) [2021-03-25T10:31:08Z] <dcaro> kernel upgrade on osds on codfw done, running performance tests (T274565)

Given that there's no noticeable improvement, will stick with the current kernel as the rest of the fleet. Might revisit
once we have metrics for osd resource efficiency.

Change 675722 had a related patch set uploaded (by David Caro; author: David Caro):
[operations/puppet@production] Revert "wmcs.ceph.codfw: Upgrade to latest 5.X kernel"

https://gerrit.wikimedia.org/r/675722

Change 675722 merged by David Caro:

[operations/puppet@production] Revert "wmcs.ceph.codfw: Upgrade to latest 5.X kernel"

https://gerrit.wikimedia.org/r/675722

Mentioned in SAL (#wikimedia-cloud) [2021-04-01T10:10:56Z] <dcaro> Restoring the 4.9 kernel on cloudcephosd2001-dev and upgrading (T274565)

Mentioned in SAL (#wikimedia-cloud) [2021-04-01T10:29:48Z] <dcaro> Done restoring the 4.9 kernel on cloudcephosd2001-dev and upgrading, requires logging into console to boot from the older kernel before removing the newer one (T274565)

Mentioned in SAL (#wikimedia-cloud) [2021-04-01T12:15:19Z] <dcaro> Restoring the 4.9 kernel on cloudcephosd2003-dev and upgrading (T274565)

Reverted all the changes on codfw, closing.