ci: gate DynamoCheckpoint deploy tests on snapshot changes only by nv-tusharma · Pull Request #11046 · ai-dynamo/dynamo

nv-tusharma · 2026-06-29T16:54:42Z

Summary

The vLLM/SGLang DynamoCheckpoint Deploy Test chain previously triggered on the operator, snapshot, or deploy filters. All three of those include the shared *ci anchor, so any CI change (and any operator/deploy change) spun up the expensive GPU vCluster checkpoint setup — running or skipping the checkpoint test even when nothing checkpoint-related changed.

This PR narrows the trigger to the snapshot filter only and drops *ci from that filter.

Changes

.github/filters.yaml: remove *ci from the snapshot filter; add tests/deploy/test_dynamocheckpoint.py so edits to the checkpoint test itself still run it.
.github/workflows/pr.yaml: narrow the gating from (operator || snapshot || deploy) to (snapshot == 'true') across all 11 checkpoint-exclusive jobs (operator setup, snapshot-agent setup, snapshot placeholders, frontend copy, the deploy tests, and cleanup) — not just the test jobs, so the expensive vCluster setup doesn't run when the test won't.

Behavior

A change under deploy/snapshot/** (or the checkpoint test file) → full checkpoint chain runs, as before.
CI-only, operator-only, or deploy-only changes → checkpoint chain is now skipped.
post-merge-ci.yml is unaffected (it has no changed-files gating; checkpoint tests still always run on merges to main).
Shared *-build / *-copy-to-acr jobs are untouched; they already include snapshot in their OR, so a snapshot-only PR still drives the full chain correctly.

🤖 Generated with Claude Code

Summary by CodeRabbit

Chores
- Updated change detection for snapshot-related CI to include an additional checkpoint test file.
- Snapshot image copy, deployment, testing, and cleanup jobs now run only when snapshot files change, reducing unnecessary pipeline runs.

The vLLM/SGLang DynamoCheckpoint deploy test chain previously triggered on the operator, snapshot, or deploy filters (all of which include the shared *ci anchor), so unrelated CI and operator/deploy changes spun up the expensive GPU vCluster setup only to run or skip the checkpoint test. Narrow the entire checkpoint-exclusive chain (operator/snapshot-agent setup, snapshot placeholders, frontend copy, the deploy tests, and cleanup) to gate on the snapshot filter only, and drop *ci from the snapshot filter so CI-only changes no longer trigger it. Add the checkpoint test file to the snapshot filter so edits to the test still run it. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

coderabbitai · 2026-06-29T16:58:14Z

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Enterprise

Run ID: 7b102c4e-7bad-4098-beef-3f9200e62897

📥 Commits

Reviewing files that changed from the base of the PR and between 1e05725 and c147f90.

📒 Files selected for processing (2)

.github/filters.yaml
.github/workflows/pr.yaml

Walkthrough

The snapshot CI filter in .github/filters.yaml drops the *ci anchor and adds tests/deploy/test_dynamocheckpoint.py. In .github/workflows/pr.yaml, all DynamoCheckpoint-related jobs (frontend copy, placeholder images, operator/agent setup, deploy tests, cleanup) have their if: conditions narrowed to trigger only on snapshot == 'true', removing the previous operator/deploy OR clauses.

Changes

Snapshot CI Filter and Workflow Gating

Layer / File(s)	Summary
Snapshot path filter `.github/filters.yaml`	Removes the `*ci` anchor from the `snapshot` block and adds `tests/deploy/test_dynamocheckpoint.py` to the tracked paths.
Narrow job if-conditions to snapshot only `.github/workflows/pr.yaml`	`frontend-copy-to-acr`, `snapshot-placeholder-vllm/sglang`, checkpoint operator/agent setup, deploy test, and cleanup jobs all drop `operator`/`deploy` from their OR conditions and gate solely on `snapshot == 'true'`.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Description check	⚠️ Warning	The description covers the change, but it misses required template sections and the mandatory Related Issues block.	Add the Overview, Details, Where should reviewer start?, and Related Issues sections, including a linked issue or a confirmed no-issue checkbox.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly summarizes the main change: gating DynamoCheckpoint deploy tests on snapshot-only changes.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{Comment @coderabbitai help to get the list of available commands.}

devin-ai-integration

Devin Review found 1 potential issue.

datadog-official · 2026-06-29T17:22:51Z

🎯 Code Coverage (details)
• Patch Coverage: 100.00%
• Overall Coverage: 41.09% (-5.96%)

_{This comment will be updated automatically if new data arrives.

🔗 Commit SHA: 0fdd367 | Docs | Give us feedback!}

galletas1712 · 2026-07-01T04:06:57Z

Can we also gate on components/src/dynamo/common/snapshot and deploy/operator/internal/checkpoint?

…dirs Address review feedback: also trigger the DynamoCheckpoint deploy tests when the shared snapshot module (components/src/dynamo/common/snapshot) or the operator's checkpoint internals (deploy/operator/internal/checkpoint) change, since both feed the checkpoint pipeline. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

nv-tusharma · 2026-07-06T20:04:36Z

Good call - added both components/src/dynamo/common/snapshot/** and deploy/operator/internal/checkpoint/** to the snapshot filter in 3aa42b0.

…t-filter # Conflicts: # .github/workflows/pr.yaml

…s only Follow the vLLM/SGLang pattern from #11046 (ci: gate DynamoCheckpoint deploy tests on snapshot changes only): narrow the changed-files gate on every TRTLLM checkpoint job (snapshot-placeholder, operator setup, snapshot-agent, deploy-test, cleanup) from (operator || snapshot || deploy) to (snapshot == 'true'). Also drop the redundant operator.result == 'skipped' branch so the operator-setup gate matches vLLM/SGLang. The snapshot path filter (.github/filters.yaml) already covers tests/deploy/test_dynamocheckpoint.py, so TRTLLM test changes still trigger these jobs. post-merge-ci.yml already gates all three frameworks identically and needs no change. Signed-off-by: Schwinn Saereesitthipitak <schwinns@nvidia.com>

nv-tusharma requested a review from a team as a code owner June 29, 2026 16:54

pull-request-size Bot added the size/M label Jun 29, 2026

github-actions Bot added ci Issues/PRs that reference CI build/test actions labels Jun 29, 2026

devin-ai-integration Bot reviewed Jun 29, 2026

View reviewed changes

Comment thread .github/filters.yaml

Merge branch 'main' into ci-checkpoint-snapshot-filter

e73189a

copy-pr-bot Bot temporarily deployed to GITLAB June 29, 2026 17:29 Inactive

copy-pr-bot Bot temporarily deployed to GITLAB June 29, 2026 17:38 Inactive

nv-anants approved these changes Jun 29, 2026

View reviewed changes

copy-pr-bot Bot temporarily deployed to GITLAB July 6, 2026 20:04 Inactive

Merge remote-tracking branch 'origin/main' into ci-checkpoint-snapsho…

c68d849

…t-filter # Conflicts: # .github/workflows/pr.yaml

galletas1712 approved these changes Jul 6, 2026

View reviewed changes

Merge branch 'main' into ci-checkpoint-snapshot-filter

0fdd367

nv-tusharma enabled auto-merge (squash) July 6, 2026 20:43

nv-tusharma merged commit 4824974 into main Jul 6, 2026
98 checks passed

nv-tusharma deleted the ci-checkpoint-snapshot-filter branch July 6, 2026 22:56

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

ci: gate DynamoCheckpoint deploy tests on snapshot changes only#11046

ci: gate DynamoCheckpoint deploy tests on snapshot changes only#11046
nv-tusharma merged 5 commits into
mainfrom
ci-checkpoint-snapshot-filter

nv-tusharma commented Jun 29, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented Jun 29, 2026

❌ Failed checks (1 warning)

Uh oh!

devin-ai-integration Bot left a comment

Uh oh!

Uh oh!

datadog-official Bot commented Jun 29, 2026 •

edited

Loading

Uh oh!

galletas1712 commented Jul 1, 2026

Uh oh!

nv-tusharma commented Jul 6, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

nv-tusharma commented Jun 29, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Behavior

Summary by CodeRabbit

Uh oh!

coderabbitai Bot commented Jun 29, 2026

Walkthrough

Changes

Estimated code review effort

❌ Failed checks (1 warning)

Uh oh!

devin-ai-integration Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

datadog-official Bot commented Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

galletas1712 commented Jul 1, 2026

Uh oh!

nv-tusharma commented Jul 6, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

nv-tusharma commented Jun 29, 2026 •

edited by coderabbitai Bot

Loading

datadog-official Bot commented Jun 29, 2026 •

edited

Loading