Due to bug#1341023, in order to change CL to 3.6, all the VMs are required to be shut down. This is a different behavior from before, that's why we need to have this requirement clearly document for the customers. Where should we document this? I think this chapter should mention it: https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.6/html/Upgrade_Guide/Upgrading_a_Red_Hat_Enterprise_Linux_6_Cluster_to_Red_Hat_Enterprise_Linux_7.html Right now I didn't find anywhere where we mention that cluster compatibility mode should be changed at all. We should mention it and we should mention also the requirement that the VMs should be shutdown. + mention the reason for this as described in the doc text in the original bug.
As an admin I strongly disagree with the "solution" that all VMs be required to be shut down for a cluster upgrade. In many (most?) real-world clusters it is neither operationally nor politically feasible to shut down all VMs in the cluster at the same time! Further, it wouldn't work as a solution anyway; the HE VM _can't_ be shut down and have the cluster still be in a state where you could change the CV (outside of manually manipulating the DB directly). Please reconsider this! Thanks, Carl
(In reply to Carl Thompson from comment #2) > As an admin I strongly disagree with the "solution" that all VMs be required > to be shut down for a cluster upgrade. It's not a question of a solution or a fix. This is an inherent property of using a "cluster level" to define a per-release set of features and functionalities applicable to VMs running in such a cluster > In many (most?) real-world clusters > it is neither operationally nor politically feasible to shut down all VMs in > the cluster at the same time! Right, and for that a bug 1348907 was implemented. It's not a perfect fix but unfortunately it's the best we can do in 3.6 as the proper way (to support VMs with different set of features within the same cluster) is missing infrastructure to do that. It will be possible from 4.0 onwards. > Further, it wouldn't work as a solution > anyway; the HE VM _can't_ be shut down and have the cluster still be in a > state where you could change the CV (outside of manually manipulating the DB > directly). Please reconsider this! Bug 1351533 is specifically about Hosted Engine > > Thanks, > Carl
(In reply to Marina from comment #0) > > Right now I didn't find anywhere where we mention that cluster compatibility > mode should be changed at all. We should mention it and we should mention it is described in https://access.redhat.com/documentation/en-US/Red_Hat_Enterprise_Virtualization/3.6/html/Upgrade_Guide/chap-Post-Upgrade_Tasks.html#Changing_the_Cluster_Compatibility_Version but without going into much details about what it actually means, that you e.g. can't resume your snapshots with memory done in previous CL, etc.
*** This bug has been marked as a duplicate of bug 1348907 ***