Skip to content

Make atelet.Terminate idempotent and write the sandbox record before Restore calls ateom - #1884

Closed
Luiz Oliveira (laoj2) wants to merge 1 commit into
agent-substrate:mainfrom
laoj2:idempotent-atelet
Closed

Luiz Oliveira (laoj2) wants to merge 1 commit into
agent-substrate:mainfrom
laoj2:idempotent-atelet

Conversation

@laoj2

Copy link
Copy Markdown
Collaborator
  • We need to write the sandbox record before Restore calls ateom, otherwise a partially failed Restore can leave an orphaned sandbox that can't be cleaned up by Terminate
  • Terminate now only skips the terminateWorkload call to ateom if it doesn't found the sandbox record, but still deregister system volumes, unmount external volumes and cleans up the snapshot and actor dirs.

@laoj2

Copy link
Copy Markdown
Collaborator Author

Talked to Zoe Zhao (@zoez7) and shrutiyam-glitch. Instead of handling failed reads to SandboxRecord, we'll send the sandbox assets in the requests to atelet.Terminate and reject the request if sandbox_assets is unset.

So, we won't need this PR

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

area/node area/reliability kind/bug Something isn't working / bugfixes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants