Summary
The push_repo_memory job fails with spawnSync git ENOBUFS on large repositories (e.g., Azure/azure-sdk-for-js with 10K+ files). The script disables sparse checkout before creating an orphan branch, which causes git to process the entire repo tree and exhaust the buffer.
Error
Disabling sparse checkout...
Checking out branch: memory/sentinel...
Branch memory/sentinel does not exist, creating orphan branch...
Switched to a new branch 'memory/sentinel'
##[error]Git command failed with error: spawnSync git ENOBUFS
Context
- The review agent itself completes successfully and posts its review
- Only the memory storage step fails
- This blocks CI status on the PR even though it's a non-critical post-processing step
Suggestion
The orphan branch creation shouldn't require disabling sparse checkout — git checkout --orphan + git rm -rf . + adding only the memory files would avoid needing the full tree.
Failing run
https://github.com/Azure/azure-sdk-for-js/actions/runs/23575363181/job/68646977344
Environment
- gh-aw v0.64.0
- Repository: Azure/azure-sdk-for-js (~10K files)
Summary
The
push_repo_memoryjob fails withspawnSync git ENOBUFSon large repositories (e.g., Azure/azure-sdk-for-js with 10K+ files). The script disables sparse checkout before creating an orphan branch, which causes git to process the entire repo tree and exhaust the buffer.Error
Context
Suggestion
The orphan branch creation shouldn't require disabling sparse checkout —
git checkout --orphan+git rm -rf .+ adding only the memory files would avoid needing the full tree.Failing run
https://github.com/Azure/azure-sdk-for-js/actions/runs/23575363181/job/68646977344
Environment