Is there documentation on how to migrate a search head deployer and a cluster master to new servers?

[a212830]

Hi,

I've been informed that my existing search-head deployer and cluster master (two different servers) need to get moved to new servers. I can't find any doc on how to do this procedure. Has anyone done it?

[Steve_G_]

To replace the deployer, it should not be necessary to grab the runtime config from a cluster member, as the previous respondent was recommending. That is, unless you are also replacing the cluster members, but that does not seem to be your need.

Rather, you merely need to configure the new deployer (to use the cluster's secret key and to set the SHC label), move the configuration bundle from the old deployer to the new deployer, and then point the cluster members to the new deployer.

The configuration steps are outlined here: http://docs.splunk.com/Documentation/Splunk/6.6.3/DistSearch/PropagateSHCconfigurationchanges#Config...

[sloshburch]

Oh nice! So is it correct that if you change the Deployer the Members won't act like "woah, who dis?! lemme just dump everything I got and start fresh with this new cool deployer dude?!" (in my world, the SH members are clearly surfers).

[somesoni2]

The steps you can use to migrate deployer is this (same as replacing deployer when original one had failure)

http://docs.splunk.com/Documentation/Splunk/6.6.3/DistSearch/PropagateSHCconfigurationchanges#How_to...

[sloshburch]

Looks like it's [at least partially] documented:

• Replace the master node on the indexer cluster

In regards to a SHC, I don't see anything explicit about changing hosts. Given that the Deployer will essentially replace what's on the SHC apps I would suggest approaching the Deployer as if you were building out the SHC from scratch - except you can grab the runtime config from one of the search heads to act as the new deployer's config.

Some additional tips:

• Open a support ticket on this activity so if you run into problems, you have a rapid channel in where they already have your 'before transition' diags and so forth
• Practice in a lab! duh!
• If not already set up, this is a great time to switch from hard coding config of the deployer and master node hostnames/ips to DNS or VIP entries thereby creating a layer of abstraction to protect if you ever have to rebuild again (or if you have a failure of one of those utility instances). In those scenarios, you can rebuild the new one and simply change the host the VIP or DNS entry target without having to touch config on each client.