Online upgrade to account for terminating pods #633
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The controller was not handling terminating pods correctly. This could lead to problems during online upgrades, such as not properly restarting primaries before taking down secondaries. The issue occurs when the operator deletes pods with the old image. If the pod is still running immediately after the deletion, the operator may skip the restart and move on to the next phase of the upgrade. This is a small window because the termination grace period is 0 seconds, but we still see it occasionally in the e2e runs. To test this issue, I added a new annotation (vertica.com/termination-grace-period-seconds) that makes it easier to hit this timing scenario. This annotation causes the pod to be in the termination step for a longer period of time. The solution was to check for pod termination in podfacts and treat the pod as not running if it is set.
roypaulin
reviewed
Dec 13, 2023
pkg/controllers/vdb/podfacts.go
Outdated
Comment on lines
72
to
73
| // true means the pod has been bound to a node, not in the process of being | ||
| // process, and all of the containers have been created. At least one |
There was a problem hiding this comment.
This sentence does not read well.
| @@ -140,6 +140,9 @@ const ( | |||
| // annotation is not provided the default value "dbadmin" will be used. | |||
| SuperuserNameAnnotation = "vertica.com/superuser-name" | |||
| SuperuserNameDefaultValue = "dbadmin" | |||
|
|
|||
| // Annotation to control the termination grace period for vertica pods. | |||
| TerminationGracePeriodSecondsAnnotaton = "vertica.com/termination-grace-period-seconds" | |||
There was a problem hiding this comment.
Is there any use for this annotation other than reproducing that issue?
There was a problem hiding this comment.
Yes, only for test purposes. There are no plans to externalize this annotation.
roypaulin
approved these changes
Dec 14, 2023
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The controller was not handling terminating pods correctly. This could lead to problems during online upgrades, such as not properly restarting primaries before taking down secondaries.
The issue occurs when the operator deletes pods with the old image. If the pod is still running immediately after the deletion, the operator may skip the restart and move on to the next phase of the upgrade. This is a small window because the termination grace period is 0 seconds, but we still see it occasionally in the e2e runs.
To test this issue, I added a new annotation (vertica.com/termination-grace-period-seconds) that makes it easier to hit this timing scenario. This annotation causes the pod to be in the termination step for a longer period of time.
The solution was to check for pod termination in podfacts and treat the pod as not running if it is set.