Set "node.kubernetes.io/exclude-from-external-load-balancers" label at reboot and unset it at startup#2948
Set "node.kubernetes.io/exclude-from-external-load-balancers" label at reboot and unset it at startup#2948ricky-rav wants to merge 1 commit intoopenshift:masterfrom
Conversation
c19ecd5 to
c07f323
Compare
dee07c1 to
e6446d1
Compare
kikisdeliveryservice
left a comment
There was a problem hiding this comment.
A few comments on a first pass
59adb63 to
03bb0cb
Compare
0346ed7 to
a659e97
Compare
kikisdeliveryservice
left a comment
There was a problem hiding this comment.
+1 on deads2k comments, but also there are 3 funcs seemingly called the same/similar things, can we condense down to 1-2 for clarity and also add in text description of funcs?
|
Latest change fixed yesterday's problem. Also noting that vsphere-upgrade did in fact pass (tho not the reqd agnostic upgrade), so this is probably ok with some retesting. /retest-required |
|
Just to followup, is this PR still needed? |
|
@kikisdeliveryservice @ricky-rav This PR would certainly be welcome; I think it would be a significant improvement in reliability. |
0392a52 to
b931d0e
Compare
|
/test unit |
|
@ricky-rav: No Bugzilla bug is referenced in the title of this pull request. DetailsIn response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
|
Issues go stale after 90d of inactivity. Mark the issue as fresh by commenting If this issue is safe to close now please do so with /lifecycle stale |
b931d0e to
a793bc9
Compare
|
/test e2e-gcp-ovn |
|
@ricky-rav: The specified target(s) for
The following commands are available to trigger optional jobs:
Use
DetailsIn response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: kikisdeliveryservice, ricky-rav The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: kikisdeliveryservice, ricky-rav The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
…it at startup Add label "node.kubernetes.io/exclude-from-external-load-balancers" on nodes that are going for a reboot and remove it at startup. This label is used by the cloud service controller in order to know which nodes to exclude from backend pools of cloud load balancers. By setting this label at very start of the reboot process, instead of waiting for the node to become NotReady, we give the cloud service controller more time to update backend pools on cloud load balancers. Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>
a793bc9 to
d8787e0
Compare
|
@ricky-rav: The following tests failed, say
Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
|
Stale issues rot after 30d of inactivity. Mark the issue as fresh by commenting If this issue is safe to close now please do so with /lifecycle rotten |
Add label
node.kubernetes.io/exclude-from-external-load-balancerson nodes that are going for a reboot and remove it at startup. This label is used by the cloud service controller in order to know which nodes to exclude from backend pools of cloud load balancers.By adding this label at very start of the reboot process, instead of just waiting for the node to become NotReady, we give the cloud service controller more time to update backend pools on cloud load balancers.
This is part of the solution for BZ #2040715