You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Paulo Motta (JIRA)" <ji...@apache.org> on 2017/11/30 17:19:00 UTC
[jira] [Commented] (CASSANDRA-14084) Disks can be imbalanced during
replace of same address when using JBOD
[ https://issues.apache.org/jira/browse/CASSANDRA-14084?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16272977#comment-16272977 ]
Paulo Motta commented on CASSANDRA-14084:
-----------------------------------------
This situation is reproduced by [this dest|https://github.com/pauloricardomg/cassandra-dtest/commit/1b96dfd855d1b2fc10cbb4cf2e4c95d236ecd951#diff-1ef92939c7765f8c4041bada71208eebR51].
The simple fix is to use normal tokens for replacement nodes with the same address:
* [3.11|https://github.com/pauloricardomg/cassandra/tree/3.11-14084]
CI looked clean when this was in CASSANDRA-13948, but I will submit again just to make sure this will not cause problems when committed separately.
> Disks can be imbalanced during replace of same address when using JBOD
> ----------------------------------------------------------------------
>
> Key: CASSANDRA-14084
> URL: https://issues.apache.org/jira/browse/CASSANDRA-14084
> Project: Cassandra
> Issue Type: Bug
> Reporter: Paulo Motta
> Assignee: Paulo Motta
>
> While investigating CASSANDRA-14083, I noticed that [we use the pending ranges to calculate the disk boundaries|https://github.com/apache/cassandra/blob/41904684bb5509595d11f008d0851c7ce625e020/src/java/org/apache/cassandra/db/DiskBoundaryManager.java#L91] when the node is bootstrapping.
> The problem is that when the node is replacing a node with the same address, it [sets itself as normal locally|https://github.com/apache/cassandra/blob/41904684bb5509595d11f008d0851c7ce625e020/src/java/org/apache/cassandra/service/StorageService.java#L1449] (for other unrelated reasons), so the local ranges will be null and consequently the disk boundaries will be null. This will cause the sstables to be randomly spread across disks potentially causing imbalance.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org