feat: add pod clique pod index to pod names by steved · Pull Request #634 · ai-dynamo/grove

steved · 2026-05-28T17:17:53Z

What type of PR is this?

/kind feature

What this PR does / why we need it:

For better at-a-glance visibility, we'd like to encode the PCLQ pod index in the pod name.

Which issue(s) this PR fixes:

Fixes #635

Special notes for your reviewer:

This moves the pod name generation out of GenerateName and into Grove. ~~Rather than appending a 5-char suffix, we can rely on the randomness of 3 characters since pod names can only conflict with their same index.~~
This tightens the validation to check names against replicas. Previously, if a 45 char name had many replicas it could exceed the hostname limit of 63 characters.
- A PCS that was previously valid may now be invalid, but that should be theoretically impossible since the pod hostnames are invalid.
Pod names can now go up to 69 chars, which is allowed (though discouraged), since we're explicitly setting the hostname and subdomain.
Note: there is currently no validation on /scale API calls

Does this PR introduce a API change?

Pod names now contain the pod clique pod index and may exceed the 63-character DNS limit

Additional documentation e.g., enhancement proposals, usage docs, etc.:

N/A

renormalize

Thanks for the PR! A few comments from me.

Issue ai-dynamo#635

renormalize

Thanks for addressing the comments @steved!

The PR looks good now - just a couple nits. We can push this in once we have more approvals from reviewers.

steved · 2026-06-25T22:41:58Z

Thanks for addressing the comments @steved!

The PR looks good now - just a couple nits. We can push this in once we have more approvals from reviewers.

Thanks for taking a look @renormalize!

unmarshall · 2026-06-02T07:44:38Z

-		if err := validatePodNameConstraints(v.pcs.Name, pcsgName, pclqName); err != nil {
+		maxPCLQReplicas := int32(1)
+		if templateSpec := componentutils.FindPodCliqueTemplateSpecByName(v.pcs, pclqName); templateSpec != nil {
+			maxPCLQReplicas = templateSpec.Spec.MaxReplicas()


If HPA is not used but lets say dynamo planner is used, then ScaleConfig is not set. This means that at best you will take template replicas. This is not an accurate representation of real time replicas since PCLQ and PCSG have their own Scale subresources and any scale-out will directly change the PCLQ and PCSG replicas and will never be reflected in PCS. So at best this check is only valuable when you solely look at PCS spec (using templates) - but with scale-outs this no longer provides any guard.

Therefore MaxReplicas name for the function is highly misleading - as it gives an impression that it always returns an accurate value during the lifetime of the PCS but that is not really true.
On top of that the doc string:

// MaxReplicas returns the maximum number of replicas for this PodClique.

is also very misleading.

unmarshall · 2026-06-02T07:49:16Z

+			maxPCLQReplicas = templateSpec.Spec.MaxReplicas()
+		}
+
+		if err := validatePodNameConstraints(v.pcs.Name, v.pcs.Spec.Replicas, scalingGroupConfig.Name, scalingGroupConfig.MaxReplicas(), pclqName, maxPCLQReplicas); err != nil {


If you wish to keep this then clearly document that this is best-effort and pre-flight and does not reflect the current state of PCS and its constituent PCLQs and PCSGs.

unmarshall · 2026-06-02T08:07:37Z

+
+func validatePodResourceClaimReferenceNames(ownerName string, maxReplicas int32, refs []grovecorev1alpha1.ResourceSharingSpec, fldPath *field.Path) field.ErrorList {
+	allErrs := field.ErrorList{}
+	maxReplicaIndex := int(max(0, maxReplicas-1))


maxReplicas is not the correct representation of what the actual max is. This value is solely derived from template spec.

unmarshall · 2026-06-27T03:45:31Z

 }

+// MaxAllowedReplicas returns the maximum number of replicas that can be configured for this PodClique.
+func (p *PodCliqueSpec) MaxAllowedReplicas() int32 {


Grove allows external pluggable horizontal autoscalers. ScaleConfig is only applicable when you wish to use HPA, otherwise this is not set (is nil). Dynamo uses planner to horizontally scale replicas. So in this case the max allowed will be limited by spec.Replicas which will not be correct.

Just so I understand - theoretically, this is still accurate, but because we're not validating the PCLQ and /scale admission we're missing changes that are applied directly? Would you recommend we add validation to those admission paths?

Bump @unmarshall if you have any thoughts, thanks!

unmarshall · 2026-06-27T03:47:10Z

 }

+// MaxAllowedReplicas returns the maximum number of replicas that can be configured with this scaling group config.
+func (p *PodCliqueScalingGroupConfig) MaxAllowedReplicas() int32 {


Please see comment

steved force-pushed the steved/pod-name-pod-index branch 2 times, most recently from ab1c617 to d161433 Compare May 30, 2026 02:41

steved changed the title ~~feat: add pod index to pod name prefix~~ feat: add pod clique index to created pod names May 30, 2026

steved force-pushed the steved/pod-name-pod-index branch 2 times, most recently from 814344e to 02d199a Compare May 30, 2026 02:44

steved changed the title ~~feat: add pod clique index to created pod names~~ feat: add pod clique pod index to pod names May 30, 2026

steved marked this pull request as ready for review May 30, 2026 02:44

steved requested review from Ronkahn21, danbar2, gflarity, sanjaychatterjee, shayasoolin and unmarshall as code owners May 30, 2026 02:44

danbar2 reviewed May 31, 2026

View reviewed changes

Comment thread operator/internal/controller/podclique/register.go

danbar2 reviewed May 31, 2026

View reviewed changes

Comment thread operator/internal/webhook/admission/pcs/validation/podcliqueset.go Outdated

yankay reviewed Jun 1, 2026

View reviewed changes

Comment thread operator/api/core/v1alpha1/podclique.go

renormalize self-assigned this Jun 2, 2026

renormalize reviewed Jun 8, 2026

View reviewed changes

renormalize removed their assignment Jun 8, 2026

yankay mentioned this pull request Jun 10, 2026

Discussion: practical name budgets for Grove-generated resources #660

Open

renormalize mentioned this pull request Jun 11, 2026

fix: include pod clique pod index in pod name for better visibility #656

Draft

steved added 2 commits June 24, 2026 11:01

feat: add pod clique pod index to pod names

83adce8

Issue ai-dynamo#635

chore: address PR feedback

9a14940

steved force-pushed the steved/pod-name-pod-index branch from 02d199a to 9a14940 Compare June 24, 2026 19:05

renormalize previously approved these changes Jun 25, 2026

View reviewed changes

Comment thread operator/internal/webhook/admission/pcs/validation/podcliqueset.go

Comment thread operator/internal/webhook/admission/pcs/validation/podcliqueset.go Outdated

chore: address PR comments

d4715ca

steved dismissed renormalize’s stale review via d4715ca June 25, 2026 22:41

chore: address PR comments

7241651

unmarshall reviewed Jun 26, 2026

View reviewed changes

unmarshall requested changes Jun 27, 2026

View reviewed changes

Uh oh!

Conversation

steved commented May 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What type of PR is this?

What this PR does / why we need it:

Which issue(s) this PR fixes:

Special notes for your reviewer:

Does this PR introduce a API change?

Additional documentation e.g., enhancement proposals, usage docs, etc.:

Uh oh!

Uh oh!

Uh oh!

Uh oh!

renormalize left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

renormalize left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

steved commented Jun 25, 2026

Uh oh!

unmarshall Jun 2, 2026

Choose a reason for hiding this comment

Uh oh!

unmarshall Jun 2, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

unmarshall Jun 2, 2026

Choose a reason for hiding this comment

Uh oh!

unmarshall Jun 27, 2026

Choose a reason for hiding this comment

Uh oh!

steved Jun 29, 2026

Choose a reason for hiding this comment

Uh oh!

steved Jul 9, 2026

Choose a reason for hiding this comment

Uh oh!

unmarshall Jun 27, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

steved commented May 28, 2026 •

edited

Loading