[
https://issues.apache.org/jira/browse/PHOENIX-7111?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17793867#comment-17793867
]
ASF GitHub Bot commented on PHOENIX-7111:
-----------------------------------------
palashc commented on PR #1744:
URL: https://github.com/apache/phoenix/pull/1744#issuecomment-1843409605
@shahrs87 I made the suggested name changes and added a test.
Something I noticed when writing a test for the server side metrics - When
we create an index on a table without async, we do 2 cache invalidations - one
to invalidate the table entry and one to invalidate index entry because index
state is updated. Could this be a problem for index creation in prod where we
might encounter errors interacting with 100s of region servers? Maybe the
second invalidation is not needed when a new index is created since none of the
region servers know about it yet, or can we piggy back the table invalidation
on the index invalidation?
> Metrics for server-side metadata cache
> --------------------------------------
>
> Key: PHOENIX-7111
> URL: https://issues.apache.org/jira/browse/PHOENIX-7111
> Project: Phoenix
> Issue Type: Sub-task
> Reporter: Palash Chauhan
> Assignee: Palash Chauhan
> Priority: Major
>
> Add metrics for monitoring the new metadata caching design.
> # Time taken to invalidate cache on all region servers during a DDL operation
> # Number of failed DDL operations due to failure in cache invalidation
> # Number of validateLastDDLTimestamp requests per region-server
> # Cache hits/misses per region-server when validating timestamps
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
