You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
To get more than static remote debugging I'm wondering if, when debugging, wrapping training commands in a screen session would allow you to then connect to a failed container that has dropped into the debugger. Potentially you could connect to multiple nodes in this way in parallel and be able to use the debugger to investigate issues with distribution?
This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.
What if anything can we do to better support TfDebugger?
Can we support interactive debugging (perhaps via kubectl exec)?
The text was updated successfully, but these errors were encountered: