Vouch request: areebahmeddd

[areebahmeddd]

What do you want to work on?

#4878

Possible approach is to handle a broken wait stream separately from a real container exit. Retry a few times and check the container before resuming monitoring.

Why this change?

A routine Docker daemon restart shouldn't kill sandboxes that survived it. Right now it leaves them stuck in Error, and from the issue even a gateway restart doesn't bring them back.

Checklist

• I wrote this request myself (not AI-generated)
• I have read CONTRIBUTING.md