On Sun, Aug 21, 2011 at 10:09 PM, Yan Chunlu <springri...@gmail.com> wrote:
> I have encountered this problem while update the key cache and row cache. > I once updated them to "0"(disable) while node2 was not available, when it > comeback they eventually have the same schema version. > > [default@prjspace] describe cluster; > Cluster Information: > Snitch: org.apache.cassandra.locator.SimpleSnitch > Partitioner: org.apache.cassandra.dht.RandomPartitioner > Schema versions: > 79d072cc-cc62-11e0-a753-5525ca993302: [node3, node1, node2] > > > > ever time I change the row cache && key cache back per CF, it show the > following error: > > [default@prjspace] update column family CommentCache with > keys_cached=200000; > > > Waiting for schema agreement... > Warning: unreachable nodes node3The schema has not settled in 10 seconds; > further migrations are ill-advised until it does. > Versions are > f7f24ef4-caf7-11e0-9b1d-5525ca993302:[node1],79d072cc-cc62-11e0-a753-5525ca993302:[node2],UNREACHABLE:[node3] > > > really have no idea what does this means.... > > > > Unreachable can mean that the node is still completing the schema migration. Only one operation can happen in this stage at a time. Try: [default@unknown] describe cluster; Might say that a node is unreachable. That can mean that node has a thread in the migration stage. You can check like so. # /usr/local/cassandra/bin/nodetool -h cdbsd03 -p 8585 tpstats | grep Migration On drop column family or truncate operations I have seen nodes stay in UNKNOWN state for a while. If they continue having trouble try a restart. If they still have trouble follow the FAQ's advice about correcting schema disagreement. Edward
