Hello,
Is there a documented process on how to upgrade a Distributed Monitoring Setup when performing a minor or major release upgrade?
The documentation states slave sites are generally only allowed to differ in patch level (when running).
Question 1:
When only upgrading the patch level, say, 1.6.0p12 to 1.6.0p27, can you upgrade the sites, master first, as per the normal upgrade instructions?
Question 2:
How would one go about minor or major version upgrades?
When we last upgraded our multisite-setup (with distributed wato) from v1.2.8 to 1.6.0, we went with a very complicated process that we figured out ourselves, and that had the advantage that it could be tested in advance (and also included a distribution upgrade or rather replacement of the OS).
But i think there must be a simpler way
We basically
1.built new servers with the new version installed and matching empty sites created in parallel
2. turned off the old sites and servers
3. restored backups of their respective in $site/var (as well as - for the master - $site/etc/check_mk, except for the apache config, which is not upgradable using cmk-update-config? ) onto the matching new servers/sites
4. ran cmk-update-config on the new sites (also needed on slaves, as it also rewrote the autocheck files)
5. disabled the slaves in the restored master config file
6. omd config ( these settings we did not restore, because the files differed)
7. generated core configs and started the sites (master first)
8. reconnected the slaves, now using TLS
9. fixed broken checks in local/share and reinventroized where necessary.
This mostly worked, and allowed us to replace the old servers rather than upgrade them - as well as test the switch in advance.
But if we want to just upgrade cmk from 1.6 to 2.0, not the underlying server, we think this may be overkill.
So, Question 2, i guess, again: Is there any documentation on how to upgrade a multisite setup? Could not find any.