hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chinmay Kulkarni (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-17959) Canary timeout should be configurable on a per-table basis
Date Wed, 17 May 2017 23:59:04 GMT

    [ https://issues.apache.org/jira/browse/HBASE-17959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16014961#comment-16014961

Chinmay Kulkarni commented on HBASE-17959:

Thanks for your comments! I have a follow-up question:
I chose to use a   _HashMap<String, AtomicLong>_  instead of a  _ConcurrentHashMap_
 since we don't really need hashmap bucket-level locking, as we are only concurrently modifying
the values corresponding to keys in the hashmap. 
Using a _ConcurrentHashMap_ could lead to an unnecessarily increased locking granularity since
multiple tables (the String keys) could be in the same bucket. Synchronizing access to the
whole map itself would lead to even more lock contention.
What are your views on this?

I will change the logging message for the actual and configured timeouts.

> Canary timeout should be configurable on a per-table basis
> ----------------------------------------------------------
>                 Key: HBASE-17959
>                 URL: https://issues.apache.org/jira/browse/HBASE-17959
>             Project: HBase
>          Issue Type: Improvement
>          Components: canary
>            Reporter: Andrew Purtell
>            Assignee: Chinmay Kulkarni
>            Priority: Minor
>         Attachments: HBASE-17959.patch
> The Canary read and write timeouts should be configurable on a per-table basis, for cases
where different tables have different latency SLAs. 

This message was sent by Atlassian JIRA

View raw message