Description of problem: Remove 3.6 from the supportedEngines reported by VDSM
supportedENGINEs = ['4.0', '4.1'] verified in vdsm-4.19.2-2.el7ev.x86_64
This is a bit confusing. the fix is in vdsm 4.19 for 4.1, but the doc text says 4.0. Should we backport this to 4.0?
There's an error in the doc text. It was required to remove 3.6 from 4.1 and 3.5 from 4.0.7 and 4.1.
Irit, is engine 4.0 then supported on vdsm 4.18?
Julio, yes, there is no reason it shouldn't be supported.
Yaniv I believe you wanted to revert this?
(In reply to Yaniv Kaul from comment #9) > Yaniv I believe you wanted to revert this? Reverting this will probably help many customers upgrading from 3.6 to 4.0 hosts directly (or even 4.1 hosts), by skipping a step of upgrading hosts to latest 3.6 vdsm and related packages.
I will set the needinfo on Yaniv back. Question to him is still relevant.
(In reply to Yaniv Kaul from comment #9) > Yaniv I believe you wanted to revert this? Yes, this is counter productive to ease of upgrades. We should add this to the testing matrix of the upgrade flow. Jiri, can you please create test cases to cover 4.x hosts in RHV 3.6 manager and 3.6 compatibility clusters?
Before the bug moves to QA and further, we should decide on its status. If the new 3.6 migration policy should be: - Upgrade your hosts to 4.0 or 4.1. (note: the meaning here is, of course, clean installation on the hosts reused from current 3.6 deployment, not actual upgrade) - Add the host to current 3.6 environment. - Migrate the manager to 4.0. - Upgrade the manager to 4.1. - Change CL level to 4.1 and done. Then this bug should not be pushed in. Yaniv/Jiri?
(In reply to Marina from comment #19) > Before the bug moves to QA and further, we should decide on its status. > > If the new 3.6 migration policy should be: > - Upgrade your hosts to 4.0 or 4.1. (note: the meaning here is, of course, > clean installation on the hosts reused from current 3.6 deployment, not > actual upgrade) > - Add the host to current 3.6 environment. > - Migrate the manager to 4.0. > - Upgrade the manager to 4.1. > - Change CL level to 4.1 and done. > > Then this bug should not be pushed in. > Yaniv/Jiri? I miss the point about "migrat(ion) the manager to 4.0/4.1" and "change cl level to 4.1...".
So, should we revert this bug?
We pushed a patch, and then reverted it. I guess the bug can stay as is.
Ok, I also see the title has changed to reflect the correct requirement. And now Jiri will update his test plan and all seem to be ok. Thanks!
*** Bug 1425198 has been marked as a duplicate of this bug. ***
ok, based on https://bugzilla.redhat.com/show_bug.cgi?id=1425198#c9
Tahlia, I am so very sorry for having wasted your time. A version with this bug was never delivered to customers, so there is no need to document it.
Hi Dan, no problem, thanks for letting me know.