[
https://issues.apache.org/jira/browse/TEZ-4097?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
László Bodor updated TEZ-4097:
------------------------------
Description:
Currently, a fetch failure is reported like this:
{code}
2019-11-05 02:50:35,972 [WARN] [Fetcher_B {Map_4} #1] |shuffle.Fetcher|: Fetch
Failure from host while connecting: *other_host*, attempt:
InputAttemptIdentifier [inputIdentifier=1, attemptNumber=0,
pathComponent=attempt_1572936153637_0005_1_00_000000_0_10003, spillType=0,
spillId=-1] Informing ShuffleManager:
java.net.SocketTimeoutException: Read timed out
...
{code}
For debugging network/ssl/etc. issues on cluster, it would be convenient to see
the local host's name in these messages (which is present in the fetcher as
localHostname property), as in the logs collected by yarn cli, it's not obvious
for the first sight.
was:
Currently, a fetch failure is reported like this:
{code}
2019-11-05 02:50:35,972 [WARN] [Fetcher_B {Map_4} #1] |shuffle.Fetcher|: Fetch
Failure from host while connecting: *other_host*, attempt:
InputAttemptIdentifier [inputIdentifier=1, attemptNumber=0,
pathComponent=attempt_1572936153637_0005_1_00_000000_0_10003, spillType=0,
spillId=-1] Informing ShuffleManager:
java.net.SocketTimeoutException: Read timed out
...
{code}
For debugging network/ssl/etc. issues on cluster, it would be convenient to see
the local host's name, which is present in the fetcher.
> Report localHostname in Fetcher failure log messages
> ----------------------------------------------------
>
> Key: TEZ-4097
> URL: https://issues.apache.org/jira/browse/TEZ-4097
> Project: Apache Tez
> Issue Type: Improvement
> Reporter: László Bodor
> Priority: Minor
>
> Currently, a fetch failure is reported like this:
> {code}
> 2019-11-05 02:50:35,972 [WARN] [Fetcher_B {Map_4} #1] |shuffle.Fetcher|:
> Fetch Failure from host while connecting: *other_host*, attempt:
> InputAttemptIdentifier [inputIdentifier=1, attemptNumber=0,
> pathComponent=attempt_1572936153637_0005_1_00_000000_0_10003, spillType=0,
> spillId=-1] Informing ShuffleManager:
> java.net.SocketTimeoutException: Read timed out
> ...
> {code}
> For debugging network/ssl/etc. issues on cluster, it would be convenient to
> see the local host's name in these messages (which is present in the fetcher
> as localHostname property), as in the logs collected by yarn cli, it's not
> obvious for the first sight.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
