julien campan created CASSANDRA-5329:
----------------------------------------
Summary: Repair -pr with vnodes incompatibilty
Key: CASSANDRA-5329
URL: https://issues.apache.org/jira/browse/CASSANDRA-5329
Project: Cassandra
Issue Type: Bug
Affects Versions: 1.2.2
Reporter: julien campan
Hi,
I have a cluster on 1.2.2 .
This cluster is composed of 16 nodes in two datacenters (8 and 8) with an RF 3
:3.
I used virtual nodes, 256 on each node.
When I do “repair –pr" on a node, I can see that it’s doing repair only on the
first vnode :
[2013-03-07 14:42:56,922] Starting repair command #7, repairing 1 ranges for
keyspace pns_fr
[2013-03-07 14:42:57,835] Repair session eb38dfa0-872c-11e2-af2d-f36fae36bab1
for range (-9064588501660224828,-9063047458265491057] finished
[2013-03-07 14:42:57,835] Repair command #7 finished
[2013-03-07 14:42:57,852] Starting repair command #8, repairing 1 ranges for
keyspace hbxtest
[2013-03-07 14:42:59,307] Repair session ebc6c7c0-872c-11e2-af2d-f36fae36bab1
for range (-9064588501660224828,-9063047458265491057] finished
So if I understand well, when I do a "repair –pr" on each node, I will repair
only the first vnode on each node. (16 token ranges on 4096 ranges).
This method doesn’t guarantee the consistency of the dataset.
It seems to me that the "repair –pr" is not compatible with vnode cluster.
--
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
