[
https://issues.apache.org/jira/browse/HADOOP-16110?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16768534#comment-16768534
]
Steve Loughran commented on HADOOP-16110:
-----------------------------------------
I see your point, but this isn't even working for the hadoop fs -ls command
For apps (e.g spark, Tez) we end up adding the relevant files to the archives
stored in HDFS, and in the cluster itself we put things somewhere on the CP
that all works on the CLI.
I'm not worried about token renewal as none of the object stores do token
renewal, even the stuff in s3a only does issuing and use in the deployed app.
Thoughts
1. I pull the env var setting up earlier
2. we cache the value of the tools env var, and then, if it has changed, warn
people that the change is ignored
for now, I've moved my env var setting into .profile, it just took me a while
to work out why things weren't getting on the CP
> ~/hadoop-env doesn't support HADOOP_OPTIONAL_TOOLS
> --------------------------------------------------
>
> Key: HADOOP-16110
> URL: https://issues.apache.org/jira/browse/HADOOP-16110
> Project: Hadoop Common
> Issue Type: Bug
> Components: bin
> Affects Versions: 3.3.0
> Reporter: Steve Loughran
> Priority: Minor
>
> if you set {{HADOOP_OPTIONAL_TOOLS}} in ~.hadoop-env, it doesn't get picked
> up because the HADOOP_OPTIONAL_TOOLS expansion takes place in the parse
> process way before {{hadoop_exec_user_hadoopenv}} is invoked.
> Unless I've really misunderstood what ~/.hadoop-env is meant to do "let me
> set hadoop env vars", I'd have expected that tools env var examining (and so:
> loading of optional tools) to take place after
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
