✨ Cached vm props / multi priority queues #1268
Merged
Conversation
This file contains hidden or 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
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
4 times, most recently
from
b17d640 to
dbfba96
Compare
github-actions
bot
added
size/XXL
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
8 times, most recently
from
d23a9c1 to
341e7b9
Compare
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
from
341e7b9 to
a576c92
Compare
bryanv
reviewed
Oct 24, 2025
|
Nice! |
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
6 times, most recently
from
70dbdcf to
02b0ebe
Compare
akutz
changed the title
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
4 times, most recently
from
523a6b1 to
f8e1a8b
Compare
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
8 times, most recently
from
499c741 to
f902e16
Compare
sbueringer
reviewed
Oct 27, 2025
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
5 times, most recently
from
06f758d to
bb4d9ef
Compare
This patch further restricts the async watcher from monitoring guest.net due to the same churn in VKS nodes that caused guest.ipStack to be so busy. This is okay since the summary.guest will still trigger a reconcile when a VM gets its primary IP. Additionally, this patch also introduces multiple priority queues for a VM's lifecycle. When a VM is going to be reconciled, it will now fall into one of the following buckets: - priorityLow int = handler.LowPriority // 0 - priorityCreating int = 100 - priorityPowerStateChange int = 90 - priorityWaitingForIP int = 90 - priorityDeleting int = 80 - priorityWaitingForDiskPromo int = 70 Thus, it no longer matters if async signal is sending thousands of VMs a minute, because VMs that are being created, deleted, waiting on a power state change, waiting for an IP, or waiting for disk promotion will all be moved to the head of the line. Finally, the VM watcher has been updated to cache VM properties for an hour and will not send async signal updates if a VM's properties have not changed. The local property cache is flushed once an hour or when the pod restarts.
akutz
force-pushed
the
feature/do-not-watch-guest-net
branch
from
bb4d9ef to
02f5a16
Compare
Minimum allowed line rate is |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
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 does this PR do, and why is it needed?
This patch further restricts the async watcher from monitoring guest.net due to the same churn in VKS nodes that caused guest.ipStack to be so busy. This is okay since the summary.guest will still trigger a reconcile when a VM gets its primary IP.
Additionally, this patch also introduces multiple priority queues for a VM's lifecycle. When a VM is going to be reconciled, it will now fall into one of the following buckets:
Thus, it no longer matters if async signal is sending thousands of VMs a minute, because VMs that are being created, deleted, waiting on a power state change, waiting for an IP, or waiting for disk promotion will all be moved to the head of the line.
Finally, the VM watcher has been updated to cache VM properties for an hour and will not send async signal updates if a VM's properties have not changed. The local property cache is flushed once an hour or when the pod restarts.
Which issue(s) is/are addressed by this PR? (optional, in
fixes #<issue number>(, fixes #<issue_number>, ...)format, will close the issue(s) when PR gets merged):Fixes
NAAre there any special notes for your reviewer:
This has been validated on a real test environment.
Please add a release note if necessary:
📚 Documentation preview 📚: https://vm-operator--1268.org.readthedocs.build/en/1268/