What is the best practice in using a local yum repository?

[Michael]

We have a local RH mirror and set up Splunk RPMs in the distro.

If a system has Splunk on it, and there's an update package in the repository, it installs the new one just fine. Except on version differences where it requires the manual start (to accept that blasted license that no one reads) and to migrate the data (well, duh, we wouldn't be updating it and not migrating the data and configs...).

Justifications aside for those silly manually required interventions on an update; yes, you can use:

splunk start --answer-yes --no-prompt --accept-license

..will get you through it, but it's still requires you to enter that line. Trying to add the switches to the script in init.d doesn't appear to be supported (or we're doing it wrong).

This can't be the first time this has been encountered, and there's much larger organizations out there than mine... So, what's the best-practice way to handle this?

Thanks!
Mike

[bishopolis]

https://www.rfaircloth.com/2017/03/07/automating-splunk-deployment-redhatcentos-poor-mans-edition/ says:

%triggerin -- splunkforwarder
/opt/splunkforwarder/bin/splunk enable boot-start --accept-license --answer-yes 
service splunk stop
mkdir -p /opt/splunkforwarder/etc/apps/org_all_deploymentclient/local    

Did you try this route and find it was no good? I'll build a similar trigger-laden package like this in a heartbeat !

[Michael · Accepted solution]

OK.

After many go rounds with Splunk support, I finally have an answer: this can't be done. I wish they would have just said that in the beginning rather than pepper me with countless links to the doc on how to run --answer-yes etc..

So, if you're reading this, with the same question: give up. Turn back.

I only have about 900 unix systems; I was hoping someone else out here would have a suggestion.

Yes, I know, Puppet...