I have a case open, but I thought I would ask here too, since time is a factor.
We have a distributed environment with several indexers, a searchhead pool for general users, and a standalone searchhead for the security team, and a deployment server. Currently the standalone searchhead is acting as the license master. We need to migrate the license master duties. Could the searchhead pool as as the license master? If not, we could migrate it to the deployment server. I need to know the best way to migrate the license master.
Thanks in advance.
I'd say use the DS as your new license master. You'll have to distribute out new master_uri settings for the licensed nodes (these are indexers, primarily, but it wouldn't hurt to add the hosts in the SHP as well.
I've done this before where I set the server.conf settings for license in an app that I then ship out via DS. Unfortunately, if your existing indexers have their license::master_uri setting in their $SPLUNK_HOME/etc/system/local/server.conf, the DS app won't work, as system/local is an override. You'll have to remove these entries manually, to allow the DS app's setting to show through.
Thanks for the reply. I forgot to mention a few other key points.
We do have an app for all license slaves. I can easily update that app and push out the changes.
Secondly, the master_uri is a DNS record that we could change the IP and all slaves would link up. We planned to use DNS names instead of IP's for this very reason.
My main concerns are with disabling the license master on the standalone, and enabling it at the new location. Pointing the slaves to the new location should not be an issue.
Apply the license to the new LM, redirect clients, when they've all checked in, remove the license from the old LM (easy in Manager -> Licensing).
@sowings IMO first we would have to check the number of deployment-client's associated with DS and their phoneHome frequency.
If its less than 50 deployment clients then we can co-host, LM and DS on same box (Since being an LM itself, is heavy duty service as each slave reports every minute). Or our box has to be really good with system resources. If all green, we can go ahead.