 Daniel Kim
|
b97c2346ae
|
Share SSH monitoring and back off transport failures (OR-558) (#741)
* Share SSH monitoring connections and persist host retry backoff
* Fix explicit reconnect, growing logs, and long SSH setup
* Require idle SSH monitors before dev-slot data cleanup
* Show SSH connection and run monitoring diagnostics
* Preserve quoted SSH known-hosts filename lists
* Include shared SSH log files in run artifact cleanup
* Support Python 3.6 remote SSH helpers
* Finish Slurm runs despite unavailable logs
* Validate SSH log chunks during concurrent rewrites
* Detect full-prefix log rewrites and isolate legacy SSH monitors
* Preserve whitespace in interactive SSH credentials
* Reuse verified log digests and ignore unrelated obsolete SSH routes
* Keep pinned SSH connections stable across unrelated config edits
* Stream large SSH command input through bounded upload frames
|
2026-10-23 14:04:56 -07:00 |
|