Github user pwendell commented on the pull request:
https://github.com/apache/spark/pull/3849#issuecomment-68406594
So personally I don't think we should change the semantics of `attemptId`
because this has been exposed to user applications and they could silently
break if we modify the meaning of the field (my original JIRA referred to an
internal use of this). What it means right now is "a global GUID over all
attempts" - that is a bit of an awkward definition, but I don't think it's fair
to call this a bug - it was just a weird definition.
So I'd be in favor of deprecating this in favor of `taskAttemptId` (a new
field) and say that it was renamed to avoid confusion. Then we can add another
field, `attemptCount` or `attemptNum` or something to convey the more intuitive
thing.
It will be slightly awkward, but if anyone reads the docs it should be
obvious. In fact, we should probably spruce up the docs here for things like
`partitionID` which right now are probably not super clear to users.
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
