[
https://issues.apache.org/jira/browse/TEZ-4511?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
László Bodor updated TEZ-4511:
------------------------------
Description:
We need to see how many times AM was notified about input read errors,
otherwise, this is a log-digging, grepping task. We might want to have it on
vertex/dag level.
I want to see 2 places:
1. Every time when an output was reported as failed to read:
https://github.com/apache/tez/blob/5beab4ced9d07bc813a8d79ded111b72af5a2f02/tez-dag/src/main/java/org/apache/tez/dag/app/dag/impl/TaskAttemptImpl.java#L1791
2. When the producer task was marked failed due to read errors and rerun. This
affects query performance even more seriously.
Basically it's when we reach this point in the code:
https://github.com/apache/tez/blob/5beab4ced9d07bc813a8d79ded111b72af5a2f02/tez-dag/src/main/java/org/apache/tez/dag/app/dag/impl/TaskAttemptImpl.java#L1888
> Counter for input read errors
> -----------------------------
>
> Key: TEZ-4511
> URL: https://issues.apache.org/jira/browse/TEZ-4511
> Project: Apache Tez
> Issue Type: Sub-task
> Reporter: László Bodor
> Priority: Major
>
> We need to see how many times AM was notified about input read errors,
> otherwise, this is a log-digging, grepping task. We might want to have it on
> vertex/dag level.
> I want to see 2 places:
> 1. Every time when an output was reported as failed to read:
> https://github.com/apache/tez/blob/5beab4ced9d07bc813a8d79ded111b72af5a2f02/tez-dag/src/main/java/org/apache/tez/dag/app/dag/impl/TaskAttemptImpl.java#L1791
> 2. When the producer task was marked failed due to read errors and rerun.
> This affects query performance even more seriously.
> Basically it's when we reach this point in the code:
> https://github.com/apache/tez/blob/5beab4ced9d07bc813a8d79ded111b72af5a2f02/tez-dag/src/main/java/org/apache/tez/dag/app/dag/impl/TaskAttemptImpl.java#L1888
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
