You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@helix.apache.org by "kishore gopalakrishna (JIRA)" <ji...@apache.org> on 2013/07/22 20:18:49 UTC

[jira] [Created] (HELIX-141) Autorebalance does not work reliably and fails when replica>1

kishore gopalakrishna created HELIX-141:
-------------------------------------------

             Summary: Autorebalance does not work reliably and fails when replica>1
                 Key: HELIX-141
                 URL: https://issues.apache.org/jira/browse/HELIX-141
             Project: Apache Helix
          Issue Type: Bug
            Reporter: kishore gopalakrishna


Autorebalance has a bug in calculating the assignment of partitions to nodes. It does not consider the fact that  a partition can have replicas. This was not tested because we generally use 1 replica ( consumer grouping). 

As we are doing this, we might want to fix/add the following features
#1. Allow user to set a minimum number of nodes to be up before computing the assignment. This minimizes the number of re-assignments when nodes start up.

#2. Have the concept of preferred node for a partition,replica. Which means when ever a partition,replica is orphaned try to first place it on its preferred location. Preferred location can be something like (p+r)%N. This will require us to use the total number of nodes to compute the preferred location. 




--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira