Github user shivaram commented on the pull request:
https://github.com/apache/spark/pull/1309#issuecomment-49807746
At a high level this is great and I think this will address my use case
that I outline in the mailing list. I do like the idea of reusing the
accumulator machinery we have instead of reinventing things.
Some questions / comments
- What happens if the accumulator is used in multiple stages ? Do we have a
start value for the accumulator when this stage starts that we will display ?
This is again a case where TaskMetrics have a defined boundary / meaning which
may be simpler to reason about than accumulators.
- Feature request: It would be great to have stats for these counters
similar to task durations etc.
- I couldn't find the class AccumulableInfo -- Does it already exist in the
tree ?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
