[
https://issues.apache.org/jira/browse/CASSANDRA-2607?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13028888#comment-13028888
]
Brandon Williams edited comment on CASSANDRA-2607 at 5/4/11 6:53 PM:
---------------------------------------------------------------------
+1
You can also automate getting the list of hosts, so long as hostname points to
the interface jmx is bound to:
{noformat}
for host in `nodetool -h \`hostname\` ring | grep -v Address | cut -f1 -d' '`;
do nodetool -h $host command; done
{noformat}
was (Author: brandon.williams):
+1
You can also automate getting the list of hosts, so long as hostname points to
the interface jmx is bound to:
{noformat}
for host in `nodetool -h \`hostname\` ring | grep -v Address | cut -f1 -d' '`;
do bin/nodetool -h $host command; done
{noformat}
> remove clustertool
> ------------------
>
> Key: CASSANDRA-2607
> URL: https://issues.apache.org/jira/browse/CASSANDRA-2607
> Project: Cassandra
> Issue Type: Task
> Components: Tools
> Reporter: Jonathan Ellis
> Priority: Minor
> Fix For: 1.0
>
>
> Clustertool is a "run an operation against the whole cluster" tool, as
> opposed to nodetool which is single-node.
> Clustertool never achieved feature parity with nodetool and never will for
> long, since it has to be manually updated for each operation type.
> Since it's trivial to just use standard tools to run nodetool against an
> entire cluster, let's drop clustertool in favor of encouraging that.
> Some examples. The first two assume a file "clustermembers" with your
> cluster host names; dsh will want that in /etc/dsh/groups instead of cwd.
> bash:
> {noformat}
> for host in `cat clustermembers`; do nodetool -h $host command;done
> {noformat}
> xargs:
> {noformat}
> # one op at a time
> cat clustermembers | xargs -n 1 -I{} nodetool -h {} command
> # parallelize
> cat clustermembers | xargs -P `wc -l clustermembers` -n 1 -I{} nodetool -h {}
> command
> {noformat}
> dsh:
> {noformat}
> # one at a time
> dsh -g clustermembers -- nodetool -h localhost command
> # parallelize
> dsh -g -c clustermembers -- nodetool -h localhost command
> {noformat}
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
