deploy: enqueue configs on pod events

[0xmichalis]

Split from #9799

This commit is valuable on its own since we get a synced config status at all times.

@mfojtik @smarterclayton

[0xmichalis]

cc: @deads2k

[deads2k]

We're updating DC status based on this somewhere?

[0xmichalis]

We're updating DC status based on this somewhere?

We now have dc.Status.Replicas (all running pods from all replication controllers), dc.Status.UpdatedReplicas (all pods from the latest replication controller), dc.Status.AvailableReplicas (all available pods), and dc.Status.UnavailableReplicas and the deployment config controller is updating those.

[deads2k]

Seems like this code won't play friendly with pod adoption by RCs owned by the DC. You should choose which information is authoritative: pods or RCs. I think that it should be RCs.

[0xmichalis]

Isn't there work happening upstream to disallow overlapping selectors?

On Wed, Jul 20, 2016 at 7:10 PM, David Eads [email protected]
wrote:

Seems like this code won't play friendly with pod adoption by RCs owned by
the DC. You should choose which information is authoritative: pods or RCs.
I think that it should be RCs.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#9953 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ADuFf07c4l3qRMFEmKLmPnW-1cWPXcJyks5qXlaUgaJpZM4JQxVB
.

[deads2k]

Isn't there work happening upstream to disallow overlapping selectors?

Adoption isn't the same as overlapping. Adoption is just picking up pods that match your selection criteria, not necessarily pods that are created by a different RC.

[0xmichalis]

Don't you want your newly adopted pods to be reflected as quickly as
possible in your DC status?

On Thu, Jul 21, 2016 at 1:49 PM, David Eads [email protected]
wrote:

Isn't there work happening upstream to disallow overlapping selectors?

Adoption isn't the same as overlapping. Adoption is just picking up pods
that match your selection criteria, not necessarily pods that are created
by a different RC.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#9953 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ADuFf7ko8RmtN209QrI4zii1E8GAB3nFks5qX1yvgaJpZM4JQxVB
.

[deads2k]

Don't you want your newly adopted pods to be reflected as quickly as
possible in your DC status?

I think that I want to have one source of truth for that information. What would you do if they disagreed? Sum of RCs gives you 10, sum of pods gives you 11.

[0xmichalis]

I think that I want to have one source of truth for that information. What would you do if they disagreed? Sum of RCs gives you 10, sum of pods gives you 11.

The thing is that RCs lack AvailableReplicas so we can only get that info from pods or unless kubernetes/kubernetes#28381 is fixed upstream.

[deads2k]

The thing is that RCs lack AvailableReplicas so we can only get that info from pods or unless kubernetes/kubernetes#28381 is fixed upstream.

Ok, can't say I'm excited about it. I'd like to come back and revisit this next release.

[merge]

[openshift-bot]

[Test]ing while waiting on the merge queue

[openshift-bot]

Evaluated for origin test up to a227c50

[0xmichalis]

Ok, can't say I'm excited about it. I'd like to come back and revisit this next release.

In my TODO list to fix the upstream issue

[openshift-bot]

continuous-integration/openshift-jenkins/test FAILURE (https://ci.openshift.redhat.com/jenkins/job/test_pr_origin/6729/)

[0xmichalis]

secrets [merge]

[0xmichalis]

yum [merge]

[0xmichalis]

#10002 [merge]

[openshift-bot]

continuous-integration/openshift-jenkins/merge SUCCESS (https://ci.openshift.redhat.com/jenkins/job/test_pr_origin/6786/) (Image: devenv-rhel7_4653)

[smarterclayton]

[merge] #9490

On Sat, Jul 23, 2016 at 8:25 AM, OpenShift Bot [email protected]
wrote:

continuous-integration/openshift-jenkins/merge FAILURE (
https://ci.openshift.redhat.com/jenkins/job/test_pr_origin/6774/)

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#9953 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/ABG_p8dex6nHtcwVmLVwd-eiDI04Kx5uks5qYgg0gaJpZM4JQxVB
.

[openshift-bot]

Evaluated for origin merge up to a227c50