[
https://issues.apache.org/jira/browse/CASSANDRA-10628?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14987958#comment-14987958
]
Robert Stupp commented on CASSANDRA-10628:
------------------------------------------
Updated the patch to pass a Java system property containing the path of the
(probably) preloaded jemalloc to log the message from a {{StartupCheck}}
implementation.
Triggered [testall + dtest for 2.2 and 3.0
branches|http://cassci.datastax.com/search/?q=snazy-10628-].
> get JEMAlloc debug output out of nodetool output
> ------------------------------------------------
>
> Key: CASSANDRA-10628
> URL: https://issues.apache.org/jira/browse/CASSANDRA-10628
> Project: Cassandra
> Issue Type: Bug
> Reporter: Jim Witschey
> Assignee: Robert Stupp
> Fix For: 2.2.4, 3.0.0
>
>
> This is a followup ticket for CASSANDRA-10581. The consensus on the Cassandra
> TE is that the debug output for loading JEMAlloc is helpful (or at least
> harmless when starting C*, but not when starting {{nodetool}}, which, it
> turns out, sources {{cassandra-env}}:
> https://github.com/apache/cassandra/blob/trunk/bin/nodetool#L53
> You can reproduce the excessive output with
> {code}
> ccm create test-nodetool-output -v git:cassandra-3.0 -n 1 ; ccm start
> --wait-for-binary-proto ; ccm node1 nodetool ring
> {code}
> The simplest solution would probably be to redirect the output of the
> sourcing to {{/dev/null/}}. I don't know if that's the right thing to do; it
> may be better to move the {{LD_PRELOAD}}-setting logic to {{bin/cassandra}}.
> I'm not sure what the reasoning would be for putting something in one place
> or another, so I leave it to the dev team to decide on the best solution.
> Assigning [~snazy] for now; feel free to reassign
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
