Joe McDonnell has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/15561 )

Change subject: IMPALA-9549: Handle catalogd startup delays when using local 
catalog
......................................................................


Patch Set 4:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/15561/4/be/src/service/impala-server.cc
File be/src/service/impala-server.cc:

http://gerrit.cloudera.org:8080/#/c/15561/4/be/src/service/impala-server.cc@2345
PS4, Line 2345:     ABORT_IF_ERROR(UpdateCatalogMetrics());
> curious to know if this means that these catalog metrics were exposed on al
Prior to this change:

Without the local catalog, coordinators have the counts right. Executors don't 
have the real counts (they often try before catalogd has started up). This is 
what executors show:
catalog.num-databases = 1
catalog.num-tables = 0

With the local catalog, it looks like both coordinators and executors get the 
right count at startup. Coordinators stay up to date. Executors do not. 
Executors also have values for other random catalog metrics beyond num-database 
and num-tables (like catalog.cache.average-load-time and others).

After this change, executors always have catalog.num-databases and 
catalog.num-tables set to zero. Coordinators stay up to date.

I will update the commit comment with this.



--
To view, visit http://gerrit.cloudera.org:8080/15561
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I1b5a94c59faaaa25927a169dcb58f310ce6b1044
Gerrit-Change-Number: 15561
Gerrit-PatchSet: 4
Gerrit-Owner: Joe McDonnell <[email protected]>
Gerrit-Reviewer: Anurag Mantripragada <[email protected]>
Gerrit-Reviewer: Impala Public Jenkins <[email protected]>
Gerrit-Reviewer: Joe McDonnell <[email protected]>
Gerrit-Reviewer: Vihang Karajgaonkar <[email protected]>
Gerrit-Comment-Date: Thu, 26 Mar 2020 22:06:08 +0000
Gerrit-HasComments: Yes

Reply via email to