Description of problem: mgr/k8sevents: Add new mgr module for kubernetes event integration New mgr module to provide a means of sending ceph related events to the kubernetes events API, and also retrieval of all kubernetes events from the rook-ceph namespace. Events may be viewed by a ceph k8sevents namespace command which will show similar output at the ceph cli to a native kubernetes client kubectl get events command. Since the events are cached by the module, it would also be possible to expose them to other mgr module(s) e.g. dashboard Version-Release number of selected component (if applicable): How reproducible: Steps to Reproduce: 1. 2. 3. Actual results: Expected results: Additional info:
Backport has been merged >1 week ago. What's the latest on this BZ?
The back-port hasn't been merged upstream, yet. In fact, it is currently in a DNM state because it is awaiting a fix for the k8sevents module. I did a partial back-port downstream to make some progress though.
(In reply to Boris Ranto from comment #5) > The back-port hasn't been merged upstream, yet. In fact, it is currently in > a DNM state because it is awaiting a fix for the k8sevents module. I did a > partial back-port downstream to make some progress though. What fix is it waiting for? Please add the issue here so we can track it in one place (https://bugzilla.redhat.com/show_bug.cgi?id=1745617 is waiting for this BZ, so I'm trying to track the whole chain of deps...)
The upstream nautilus back-port PR is in the DNM state: https://github.com/ceph/ceph/pull/30215 It is waiting for a fix for: https://tracker.ceph.com/issues/41737 i.e. module currently crashes in non-k8s environments.
(In reply to Boris Ranto from comment #8) > The upstream nautilus back-port PR is in the DNM state: > > https://github.com/ceph/ceph/pull/30215 > > It is waiting for a fix for: > > https://tracker.ceph.com/issues/41737 PR - https://github.com/ceph/ceph/pull/30482 (easier to track by PR if available) > > i.e. module currently crashes in non-k8s environments.
There was some progress upstream so I have cherry-picked the rest of the necessary commits for this bz. However, we do have an issue with our downstream builds at the moment as one of the back-ported patches (not for this bz) introduced a build failure on s390x. The s390x build failure will need to be fixed before we can move this to ON_QA.
Hi Anmol, Could you please provide steps for testing this functionality? (QE need this in order to qa_ack this BZ)
pcuzner can provide the information as he developed the feature.
The merge is still pending upstream - but if we have it downstream in our container build that's great. Ultimately the module should be enabled by rook - but until this is done, you'll need to enable manually through the tools pod. ceph mgr module enable k8sevents this registers a coupld of commands ceph k8sevents status | ls | ceph So once loaded you can use those commands to check that it's operational. When storageclasses get created, or osds get added/removed or healthcheck fails you chould see kubernetes events show up in the OCS dashboard. Raz, if you can confirm the above I'll start the process to auto enable the module within rook.
Hi Paul, Is it just for testing? I'm asking because the tools pod is not going to be part of OCS
the tools pod is not required for this. k8sevents is a mgr module, so as long as it's in the image we can enable manually to test it and then raise a PR for rook to automatically enable the module - like we do for prometheus
@Elad, can you please assign this for verification?
verified
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:0312