WL#2702: Add "cycle" command to ndb_mgm
Affects: Server-7.1 — Status: Un-Assigned — Priority: Medium
It is a pain in the @#$% to have to restart each data node by hand to upgrade or to have a new configuration setting take affect. This becomes a huge undertaking if you have 8 or more data nodes. I propose a cycle command that will restart each data node one after the other and then tell the user when done. Currently is you issue a 3 restart, you get back "Node 3 is being restarted." I would like to see the following; ndb_mgm> cycle Node 2 is being restarted. Node 2 has been restarted. Node 3 is being restarted. Node 3 has been restarted. Node 4 is being restarted. Node 4 has been restarted. Node 5 is being restarted. Node 5 has been restarted. Node 6 is being restarted. Node 6 has been restarted. Node 7 is being restarted. Node 7 has been restarted. Node 8 is being restarted. Node 8 has been restarted. cycle complete ndb_mgm> Comment added by Trudy Pelzer, 2005-10-25: Max Mether opened a duplicate entry (see WL#2945, which I have cancelled). This is his request: There should be a command in the management client for doing a rolling restart of all data nodes. Something like ndb_mgm> ROLLING|CYCLIC RESTART [i] This would do a RESTART for each data node, and then RESTART the following when it's "partner" comes back up. So that there would only be one node restarting at the same time. If a node doesn't come back up for some reason, or a node crashes durinng the command, then the command would abort and display which nodes were restarted.
Copyright (c) 2000, 2015, Oracle Corporation and/or its affiliates. All rights reserved.