You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues-all@impala.apache.org by "ASF subversion and git services (Jira)" <ji...@apache.org> on 2023/06/08 16:36:00 UTC
[jira] [Commented] (IMPALA-10794) Improve observability of execDdl code paths
[ https://issues.apache.org/jira/browse/IMPALA-10794?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17730641#comment-17730641 ]
ASF subversion and git services commented on IMPALA-10794:
----------------------------------------------------------
Commit f5aa242de6ae78bab7f0532204637797ba8b84c0 in impala's branch refs/heads/master from Michael Smith
[ https://gitbox.apache.org/repos/asf?p=impala.git;h=f5aa242de ]
IMPALA-12158: Restore process name for daemons
IMPALA-10794 introduced names for threads, which messed up process names
too for catalogd and impalad.
Sets a pseudo-property sun.java.command that is used by jps to read the
Java process name. Use that property to replace the name of the "main"
thread (which, when updated, also updates the process name).
The process name is also used for renames in ThreadNameAnnotator so that
the original process name is always present. This ensures tools like
pgrep can find impalad and catalogd.
Change-Id: I7469b4f595acc1949286520a8e6086716bec0b1f
Reviewed-on: http://gerrit.cloudera.org:8080/19990
Reviewed-by: Impala Public Jenkins <im...@cloudera.com>
Tested-by: Michael Smith <mi...@cloudera.com>
> Improve observability of execDdl code paths
> -------------------------------------------
>
> Key: IMPALA-10794
> URL: https://issues.apache.org/jira/browse/IMPALA-10794
> Project: IMPALA
> Issue Type: Improvement
> Components: Catalog, Frontend
> Reporter: Quanlong Huang
> Assignee: Quanlong Huang
> Priority: Critical
> Labels: observability, supportability
> Fix For: Impala 4.2.0, Impala 4.1.1
>
>
> It's hard to debug a hanging DDL/DML:
> * The last log before coordinator sending execDdl RPC to catalogd is "Analysis and authorization finished."
> * Catalogd just prints logs when it finishes the HMS operations and start reloading metadata incrementally.
> It'd be helpful to add logs:
> * On coordinator side, add logs when start/finish the execDdl RPCs.
> * On catalogd side, add logs when start/finish executing an execDdl request. More loggings in specifit complex ddls will also be helpful.
> We also need to add thread annotation in catalogd for the part executing HMS RPCs. We currently only have them in loading metadata. That's not enough.
> CC [~vihangk1] who may already have a draft for this.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-all-unsubscribe@impala.apache.org
For additional commands, e-mail: issues-all-help@impala.apache.org