cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Joel Segerlind <j...@kogito.se>
Subject Slow pre-decommission repair
Date Tue, 17 Dec 2013 16:00:09 GMT
Hi,

I'm trying to decommission a node from a six node Cassandra 2.0.3 cluster,
following the instructions at
http://www.datastax.com/documentation/cassandra/2.0/webhelp/index.html#cassandra/operations/ops_remove_node_t.html.
The node has just short of 11GB data in one keyspace (RF=3), and the repair
is taking forever (around 25 hours if it's keeping the current pace).

Re-reading the instructions I realized that I ran nodetool repair, instead
of nodetool repair *keyspace_name*, but this can't have this kind of impact
right? I ran nodetool repair -pr on the same node yesterday, and it took
around 35 min.

The cluster is not under significant load, neither CPU, RAM, disk nor
network IO.

Nothing has ever been deleted from this keyspace and no entries expire.

Does anyone have any idea why a repair would take this long?

When decommissioning nodes, would it be sufficient to run nodetool repair
-pr on the node to be decommissioned?

Mime
View raw message