I'd like to make sure you've seen the perform_maintenance_hosts <https://github.com/apache/incubator-aurora/blob/e6e7e53d92b52d78960824022bef8a0546002180/src/main/python/apache/aurora/client/commands/maintenance.py#L81> command, which will take care of the two-step draining <https://github.com/apache/incubator-aurora/blob/e6e7e53d92b52d78960824022bef8a0546002180/src/main/python/apache/aurora/admin/host_maintenance.py#L31> Aurora uses.
Also, Maxim is currently working on improvements to how the maintenance API is exposed in the admin clien <https://issues.apache.org/jira/browse/AURORA-43>t, which should improve the process significantly. On Thu, Aug 21, 2014 at 1:13 PM, Maxim Khutornenko <ma...@apache.org> wrote: > Hi Bhuvan, > > Looks like you are missing a cluster name argument in your command. > Something like "aurora_admin start_maintenance_hosts --hosts=<hosts> > <cluster>" should do it. > > Thanks, > Maxim > > On Thu, Aug 21, 2014 at 1:01 PM, Bhuvan Arumugam <bhu...@apache.org> > wrote: > > I'm trying to use aurora_admin command to bring few hosts in > > maintenance mode. I dont want to schedule jobs in these hosts, as they > > may be down or due to be rebooted soon. > > > > The aurora_admin interface dont seem to behave as advertised. Sounds > > like a bug with how options are parsed? > > > > [root@scheduler ~]# aurora_admin start_maintenance_hosts > --hosts="slave1,slave2" > > FATAL] usage: start_maintenance_hosts {--filename=filename | > --hosts=hosts} > > > > see the help subcommand for more details. > > > > [root@scheduler ~]# cat /tmp/maintenance.hosts > > slave1 > > [root@scheduler ~]# aurora_admin start_maintenance_hosts > > --filename=/tmp/maintenance.hosts > > FATAL] usage: start_maintenance_hosts {--filename=filename | > --hosts=hosts} > > > > see the help subcommand for more details. > > > > > > -- > > Regards, > > Bhuvan Arumugam > > www.livecipher.com >
