hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Raghu Angadi (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HBASE-1939) HLog group commit
Date Thu, 29 Oct 2009 18:37:59 GMT

    [ https://issues.apache.org/jira/browse/HBASE-1939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12771542#action_12771542
] 

Raghu Angadi commented on HBASE-1939:
-------------------------------------

Thanks J-D.

I should have seen the update earlier... not sure how important 'force' flag is. Currently
it might not work properly.

Say sync thread starts fs_sync() at time t0. At t1 a client appends and invokes a force-sync...
I am assuming the contract is  it should return after next fs_sync. But in the current implementation,
it may not : when fs_sync() returns at t2, it sets this flag to false.. client's force-sync
could return without a fs_sync.

I might implement it something like :

{code}
 addToSyncQ(boolean force) {
   lock()
   if (force) {
        this.forceSync = true;
   }
  //...
} {code}

Rest of the implementation remains same.. with forceSync being a simple boolean rather than
an atomic.


> HLog group commit
> -----------------
>
>                 Key: HBASE-1939
>                 URL: https://issues.apache.org/jira/browse/HBASE-1939
>             Project: Hadoop HBase
>          Issue Type: New Feature
>            Reporter: Jean-Daniel Cryans
>            Assignee: Jean-Daniel Cryans
>             Fix For: 0.21.0
>
>         Attachments: HBASE-1939-v2.patch, HBASE-1939.patch
>
>
> Currently multiple clients writing to a RS are serialized when sync'ing their appends.
Implementing group commit can help this by being aware of all the clients in that queue.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message