Do not check SSR State when Snapshotter should be stopped #680
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
|
@avestuk Thank you for your contribution. |
|
|
gardener-robot
added
needs/review
|
Thank you @avestuk for your contribution. Before I can start building your PR, a member of the organization must set the required label(s) {'reviewed/ok-to-test'}. Once started, you can check the build status in the PR checks section below. |
aaronfern
added
the
reviewed/ok-to-test
gardener-robot-ci-3
added
needs/ok-to-test
|
@ishan16696 Anything I can do to help move this along? |
|
Hi @avestuk , |
|
Thanks @avestuk for PR, |
|
As requested in this comment: #680 (comment), please rebase this PR on latest master. |
added 3 commits
March 25, 2024 08:23
Allows type checking of channel interactions
I think that closing the channel makes it clearer that this channels purpose has been fulfilled and it no longer needs to be considered. All the places that use the channel are recieving all values from the channel so they will continue to function in the same way
The snapshotter is not set to active state until an initial delta snapshot or a fullsnapshot has been taken. If a leadership election takes place while the snapshot is being taken and the current leader loses that election, then handleSsrRequest does not send on the ssrStopCh. This in turn causes ssr.snapshotEventHandler to never return and snapshots would continue to be taken.
avestuk
force-pushed
the
specify-channel-types
branch
from
e59dd3f to
c90aa01
Compare
Thanks very much @ishan16696 I've rebased on latest master. |
ishan16696
added
reviewed/ok-to-test
gardener-robot-ci-3
removed
the
reviewed/ok-to-test
|
/hold until #716 is merged. |
gardener-robot
added
the
reviewed/do-not-merge
ishan16696
approved these changes
Mar 26, 2024
ishan16696
removed
the
reviewed/do-not-merge
PR #716 will take some time as it also need to resolve merge conflicts, let's not block this PR. Talked to @shreyas-s-rao and he is fine with it. |
gardener-robot
added
the
status/closed
ishan16696
pushed a commit
to ishan16696/etcd-backup-restore
that referenced
this pull request
Jul 15, 2024
* Specify send/recv Allows type checking of channel interactions * Close instead of sending on channel I think that closing the channel makes it clearer that this channels purpose has been fulfilled and it no longer needs to be considered. All the places that use the channel are recieving all values from the channel so they will continue to function in the same way * Do not check ssr state in handleSsrStopRequest The snapshotter is not set to active state until an initial delta snapshot or a fullsnapshot has been taken. If a leadership election takes place while the snapshot is being taken and the current leader loses that election, then handleSsrRequest does not send on the ssrStopCh. This in turn causes ssr.snapshotEventHandler to never return and snapshots would continue to be taken.
ishan16696
added a commit
that referenced
this pull request
Jul 15, 2024
* Specify send/recv Allows type checking of channel interactions * Close instead of sending on channel I think that closing the channel makes it clearer that this channels purpose has been fulfilled and it no longer needs to be considered. All the places that use the channel are recieving all values from the channel so they will continue to function in the same way * Do not check ssr state in handleSsrStopRequest The snapshotter is not set to active state until an initial delta snapshot or a fullsnapshot has been taken. If a leadership election takes place while the snapshot is being taken and the current leader loses that election, then handleSsrRequest does not send on the ssrStopCh. This in turn causes ssr.snapshotEventHandler to never return and snapshots would continue to be taken. Co-authored-by: Alex Vest <avestuk@gmail.com>
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.
What this PR does / why we need it:
As reported in #676 we sometimes see that multiple snapshotters can be running at one time in a cluster.
I believe that the reason for this is that the snapshotter is not set to active state until an initial delta snapshot or a fullsnapshot has been taken. If a leadership election takes place while the snapshot is being taken and the current leader loses that election, then handleSsrRequest does not send on the ssrStopCh. This in turn causes ssr.snapshotEventHandler to never return and snapshots would continue to be taken.
Secondly I'm propsing a change to close the
ssrStopChrather than sending on it. This is to make it clear to the reader that the channel is no longer being used at this point.Lastly the function signature changes are simply adding further type annotations to the channels passed to functions to help readers understand where channels are being read from/sent to.
Which issue(s) this PR fixes:
Fixes #676
Special notes for your reviewer:
The reason why the ssrState was being checked is not apparent to me so I'd be delighted to be told why this might be an awful idea.
Release note: