New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
operator: remove pod list of an entire cluster #11376
Conversation
Please set the appropriate release note label. |
1eee453
to
3371814
Compare
test-me-please |
test-me-please |
operator/k8s_cep_gc.go
Outdated
// there is more data, continue | ||
continue perCEPFetch | ||
if !exists { | ||
// FIXME: this is fragile has we might have received the |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
there are a few typos in this comment...
test-me-please |
test-me-please |
retest-net-next |
ad9813c
to
842782d
Compare
test-me-please |
operator/identity/heartbeat.go
Outdated
@@ -57,6 +58,11 @@ func (i *IdentityHeartbeatStore) IsAlive(identity string) bool { | |||
i.mutex.RLock() | |||
defer i.mutex.RUnlock() | |||
|
|||
// The identity is definitely alive if there's a CE using it. | |||
if watchers.HasCEWithIdentity(identity) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This looked very expensive at first but the index makes it bearable it seems.
test-me-please |
5f1e79b
to
199c64c
Compare
test-me-please |
Doing a pod list every 30 minutes can be brutal for kube-apiserver, specially if it is not running with a limit of the number of pods that should be returned. Having a watcher that receives notifications for the pods running as well as the CEPs, is enough to provide a GC for Cilium Identities and CEPs. Since we have a k8s watcher and the CEPs have an owner set to themselves the CEP GC interval was decreased to 5 minutes. Also, given that we don't depend on the CEP GC to perform Cilium Identity GC we can set the identity heartbeat timeout to 2* the KVStore Lease TTL. Signed-off-by: André Martins <andre@cilium.io>
test-me-please |
hit #11512 |
Doing a pod list every 30 minutes can be brutal for kube-apiserver, specially if it is not running with a limit of the number of pods that should be returned. Having a watcher that receives notifications for the pods running as well as the CEPs, is enough to provide a GC for Cilium Identities and CEPs.
Since we have a k8s watcher and the CEPs have an owner set to themselves the CEP GC interval was decreased to 5 minutes.
Also, given that we don't depend on the CEP GC to perform Cilium IdentityGC we can set the identity heartbeat timeout to 2* the KVStore Lease TTL.
Signed-off-by: André Martins andre@cilium.io
Fixes #11472