[DownstreamMerge] 8-8-2022 downstream merge without ep slice #1246
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
Signed-off-by: Pardhakeswar Pacha <ppacha@nvidia.com>
Apparently if you don't pass a ResourceVersion the call goes directly through to etcd, while if you pass "0" it gets pulled from the apiserver's cache. Pulling from the cache can greatly reduce load on etcd. See: https://github.com/kubernetes/kubernetes/blob/master/staging/src/k8s.io/apiserver/pkg/storage/cacher/cacher.go#L637 // GetList implements storage.Interface func (c *Cacher) GetList(ctx context.Context, key string, opts storage.ListOptions, listObj runtime.Object) error { recursive := opts.Recursive resourceVersion := opts.ResourceVersion pred := opts.Predicate if shouldDelegateList(opts) { return c.storage.GetList(ctx, key, opts, listObj) } <...> } func shouldDelegateList(opts storage.ListOptions) bool { resourceVersion := opts.ResourceVersion pred := opts.Predicate pagingEnabled := utilfeature.DefaultFeatureGate.Enabled(features.APIListChunking) hasContinuation := pagingEnabled && len(pred.Continue) > 0 hasLimit := pagingEnabled && pred.Limit > 0 && resourceVersion != "0" // If resourceVersion is not specified, serve it from underlying // storage (for backward compatibility). If a continuation is // requested, serve it from the underlying storage as well. // Limits are only sent to storage when resourceVersion is non-zero // since the watch cache isn't able to perform continuations, and // limits are ignored when resource version is zero return resourceVersion == "" || hasContinuation || hasLimit || opts.ResourceVersionMatch == metav1.ResourceVersionMatchExact } Signed-off-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>
Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>
This denies resources that specify an invalid cidr. Signed-off-by: Ori Braunshtein <obraunsh@redhat.com>
Isolate syncMap implementation from the retry logic: - retryObjEntry doesn't need to have mutex anymore - retryObjs doesn't work with cache mutexes, only calls methods syncMap is a map with lockable keys. It allows to lock the key regardless of whether the entry for given key exists. When key is locked other threads can't read/write the entry with the key. Split ensureRetryEntryLocked into loadOrStore that will make sure newEntry is present and locked, and load that was previously called as ensureRetryEntryLocked with nil newRetryEntry. Change iterateRetryResources to only lock every entry once and not use cache mutex. Create a snapshot of keys that will be retried instead of holding map lock. Lock retryObject key until work for this key is completed. That also removes the need for .ignore field - retry loop will wait until key is unlocked. Signed-off-by: Nadia Pinaeva <npinaeva@redhat.com>
Fix retry_obj retryMutex vs retryEntry.Mutex deadlocks
kube: pass ResourceVersion:"0" for direct List() calls
Add EgressQoS DstCIDR kubebuilder validation
Signed-off-by: Girish Moodalbail <gmoodalbail@nvidia.com>
... since golangci-lint might install a later version by design: golangci/golangci-lint-action#75 Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>
Signed-off-by: Riccardo Ravaioli <rravaiol@redhat.com>
…e informer cache When processing an object in terminal state there is a chance that it was already removed from the API server. Since delete events for objects in terminal state are skipped delete it here. Signed-off-by: Patryk Diak <pdiak@redhat.com>
On update, delete objects in terminal state that no longer exist in the informer cache
Followup to EndpointSlices PR for ovn-k node
Adding and removing a pod on changing nodes back to back can end up in a race where corresponding logical switch port remains in the wrong logical switch and never gets properly removed. In order for this to happen, the logical switch port has to have the same name, which is the <namespace>_<podName>. Signed-off-by: Flavio Fernandes <flaviof@redhat.com> Co-authored-by: Tim Rozet <trozet@redhat.com>
BZ2117310: Fix race when adding and removing pod with same name
This reverts commit 503de2f.
…ube-node" This reverts commit 1a225ea.
|
/hold |
openshift-ci
bot
added
the
do-not-merge/hold
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: trozet The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
openshift-ci
bot
added
the
approved
|
/retest-required |
|
@trozet: The following tests failed, say
Full PR test history. Your PR dashboard. 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. I understand the commands that are listed here. |
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.
backup plan if #1237 fails