Compute Worker - Docker image update check added

[ihsaan-ullah]

Description

• This PR adds additional code to compute worker to compare local and remote compute worker images
• It also has small update to circle-ci config to run flake-8 on compute worker directory

Logs

Compute worker container will show logs like below when this new code runs

Issues this PR resolves

• Computer worker upgrade feature #2320

Checklist

• Code review by me
• Hand tested by me
• I'm proud of my work
• Code review by reviewer
• Hand tested by reviewer
• CircleCi tests are passing
• Ready to merge

[Didayolo]

So this adds warning in the worker logs when the version is not up-to-date?

• Do we have anything visible on the platform's UI? Like, the logs appear in the submission logs?
• Can we differentiate mandatory updates from optional updates? (that is maybe not that useful)

[ihsaan-ullah]

So this adds warning in the worker logs when the version is not up-to-date?

Yes

• Do we have anything visible on the platform's UI? Like, the logs appear in the submission logs?

No, this will only be visible in compute worker logs

• Can we differentiate mandatory updates from optional updates? (that is maybe not that useful)

For now this will only check if local is behind reemote or not. Later on if needed we can somehow differentiate between critical and non-critical updates. This first version do not interact with anything except the docker image on remote(docke hub)

[Didayolo]

I think the warning should be visible on platform side, typically in ingestion_error_logs and scoring_error_logs.

For the "optional or mandatory" update I think it is not that important.

[Didayolo]

Update:

• In case of problem we write logs in "scoring error logs" and "ingestion error logs"
• We check by current container ID instead of just the docker image

The reason for this second point is that, otherwise, if you git pull the latest image but do not restart the container, it will be marked as up-to-date.

The problem with this approach is that in many case the container won't be found by inspect_container just by container ID (typically because of the format "projectname_containername"), so it will throw an error and go back to only checking the image.