Github user squito commented on a diff in the pull request:
https://github.com/apache/spark/pull/17113#discussion_r106248521
--- Diff: docs/configuration.md ---
@@ -1411,6 +1411,15 @@ Apart from these, the following properties are also
available, and may be useful
</td>
</tr>
<tr>
+ <td><code>spark.blacklist.application.fetchFailure.enabled</code></td>
+ <td>false</td>
+ <td>
+ (Experimental) If set to "true", Spark will blacklist the executors
immediately when the fetch failure
+ happened. If external shuffle service is enabled, then the whole node
will be blacklisted. This configuration
+ is to handle some scenarios where shuffle fetch is available and
cannot be recovered through retry.
--- End diff --
the last sentence is pretty confusing ... I am not sure what else to put,
maybe just leave it out?
This change would be helpful when fetch failures come from non-transient
issues, but would cause extra recomputations when the failures are transient.
Unfortunately that doesn't really give much guidance to the end user ... I feel
like even ourselves we don't really have clear guidelines on when it should be
set, its sort of an emergency escape-hatch we're leaving in place.
---
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]
