Skip to content

Investigate why remote_monitor takes time to shutdown after experiment is over #35

Description

@milindsrivastava1997

I've observed this in long experiments with large query lookback periods. There are long multiple instances of:
ssh -o StrictHostKeyChecking=no milindsr@node0.scratch1.cloudmigration-PG0.utah.cloudlab.us "pgrep -f remote_monitor.py" Remote monitor is still running. Will check again in 10 seconds

Not a problem wrt functionality but figuring this out will speed up running of experiments.
My hypothesis was that this was because of remote_monitor and/or PrometheusClient dumping large files and taking time to do so.
However, remote_monitor's monitor_output.json is small and PrometheusClient has been updated to stream results to file instead of dumping them all at once after the end of an experiment.

Activity

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

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions