Deployment Server Reload Behavior

[dcparker]

Hi all,

I am working on putting my deployment server code in an external location, like GitHub. This part is working great. The problem comes when new code gets pushed to the deployment server, I am having to manually run the "splunk reload deploy-server." I can script that part, but I am worried about running the blanket reload when only 1 class changes in serverclass.

My question - does not specifying a class during the reload cause all forwarders to restart and look for new code or is Splunk smart enough to only tell the ones that changed to restart after getting new code? I've been doing it per class, because I thought it was bad to refresh the whole environment if only a single class changed, but this might be a bad assumption. It's also hard to script per class, especially if someone added a new class from Git (at least with the way I am thinking about it.)

Anyone have any thoughts on this?

Let me know if you need more info or don't understand the question, thanks!

[Ayn · Accepted solution]

Whether a deployment client will restart or not is determined by whether the restartSplunkd parameter is set to True or False (default is False), AND additionally if set to True this will only happen when a client actually gets an updated app from the deployment server. In other words, no restarts will happen on clients that aren't affected by any updated apps. The docs on serverclass.conf really explains this pretty well, see http://docs.splunk.com/Documentation/Splunk/5.0.2/Admin/Serverclassconf

[dcparker]

Thanks for your help!