Github user squito commented on the issue:
https://github.com/apache/spark/pull/14079
@kayousterhout @tgravescs Sorry for the long-delay on my end, I have
finally updated this. I think I have addressed all the naming / commenting
issues -- I'm going to do a bit more testing on it and then come back for a
fresh pass to see if there are more things.
Some important changes in this revisions:
* timeouts of individual task failures, in application blacklisting only.
This is to avoid blacklisting an executor in a long-lived app when you've got a
few random failures that are very far apart in time. But everything still has
its own timeout -- here's the comment I've put in the code to explain that:
```scala
// Note that we do *not* remove executors and nodes from
// the blacklist as we expire individual task failures -- each have
their own timeout. Eg.,
// suppose:
// * timeout = 10, maxFailuresPerExec = 2
// * Task 1 fails on exec 1 at time 0
// * Task 2 fails on exec 1 at time 5
// --> exec 1 is blacklisted from time 5 - 15.
// This is to simplify the implementation, as well as keep the
behavior easier to understand
// for the end user.
```
* config name changes. Also, docs mark everything as "Experimental" except
for spark.blacklist.enabled. The idea is to convey that blacklisting will
continue to be exported, but the specific knobs may go through another revision.
* bug fix -- in some more testing, I discovered that I was blacklisting
nodes whenever the *total* number of blacklisted executors was over the
threshold, rather than checking the number of blacklisted executors on *that
node*. Fixed, cleaned up code a bit, added a test.
---
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 infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
