Closed Bug 1269766 Opened 10 years ago Closed 9 years ago

Test / deploy the Spark progress bar extension on a.t.m.o

Categories

(Cloud Services Graveyard :: Metrics: Pipeline, defect, P1)

defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: mreid, Assigned: jezdez)

References

Details

Install the extension from https://github.com/mozilla/jupyter-spark and test to ensure that it works as expected on our a.t.m.o infrastructure. If so, add it to our deploy scripts. This provides a progress bar for Spark cells in notebooks, as well as offering a basic proxy to the Spark Web UI (to avoid having to configure tunnels / proxies / port forwarding).
Points: --- → 2
Priority: -- → P3
Assignee: nobody → jezdez
Status: NEW → ASSIGNED
Making this P2 since I'm not sure if I'll do this in the short sprint this week and it depends on #1281209.
Depends on: 1281209
Priority: P3 → P2
Priority: P2 → P1
I've tested it on the infrastructure after doing some code re-organization. This needs more tests, a PyPI release now and an update to the EMR repo next.
When I tested the extension on a Spark cluster I found some issues with the proxy code and had to rewrite some of it to use Tornado's async code. That just got released to PyPI as 0.3.0 (https://github.com/mozilla/jupyter-spark/compare/0.2.0...0.3.0). Code test coverage is now at 100% for the Python code. Next TODO is another round of testing on a Spark cluster and an update to the EMR repo.
Okay, manual tests on a fresh cluster were successful. I opened https://github.com/mozilla/emr-bootstrap-spark/pull/29
This has been deployed now and is available on newly created clusters.
Status: ASSIGNED → RESOLVED
Closed: 9 years ago
Resolution: --- → FIXED
Product: Cloud Services → Cloud Services Graveyard
You need to log in before you can comment on or make changes to this bug.