Description of problem: Currently it is not possible to update a Node el6 host to Node el7 host. The problem is that all host specific configuration is lost. To prevent this, a mechnism is needed to perform major version upgrades.
Rough idea: 1. Install 3.5 el6 2. Register to engine 3.6 3. Add host to 3.5 compat cluster 4. Check that everything is working 5. Put host into maintenance mode 6. On the host: Save the contents of /config (i.e. create tarball and copy to other host) 7. Reinstall host with 3.6 el7 (don't register) 8. Boot into host, unmount all files from /config 9. On the host: Empty /config, restore /config from tarball from step 6 10. Reboot host with permissive mode 11. On the host: restorecon -r / (warnings are okay) 12. Activate host in Engine 13. Check that everything is working 14. rescude audit.log
Also need to consider In the same cluster, there probably have more than one Node 6 hosts, most of the situation is this, and Node 6 and Node 7 can not be in same cluster in RHEV-M.
Yes, that also needs to be considered. Maybe it's even viable to say: 1. Start wtih all hosts in cluster A (el6) 2. For each host: Update (as in comment 1) and move to cluster B (el7)
This will be a migration, not an upgrade, thus there will be no way to roll back. However, a rollback is possibly, to reinstall with the old el6 image again, and restore the saved config there. Also: We can only do this with reinstall, because the partition sizes differ between el6 and el7.
Bug tickets must have version flags set prior to targeting them to a release. Please ask maintainer to set the correct version flags and only then set the target milestone.
Target release should be placed once a package build is known to fix a issue. Since this bug is not modified, the target version has been reset. Please use target milestone to plan a fix for a oVirt release.
Huijuan, could you give a roughly try according to comment 1 Fabian provided.
Raising priority, because the same flow might be used for vintage to NGN migration
I tried to test this issue according to comment 1, test results are same as expectation in comment 1. Test version: RHEVH-6.7-20160219.0.el6ev RHEVH-7.2-20160328.0.el7ev Test steps: 1. Install 3.5 el6: RHEVH-6.7-20160219.0.el6ev 2. Register to engine 3.6 3. Add host to 3.5 compat cluster (configure the "Management Network" to "rhevm") 4. Check that everything is working 5. Put host into maintenance mode on rhevm 6. On the host: Save the contents of /config (create config.tar.gz and copy to other host) 7. Reinstall host with 3.6 el7:RHEVH-7.2-20160328.0.el7ev (did not register) 8. Boot into host, unmount all files from /config (rm all files from /config) 9. On the host: Empty /config, restore /config from tarball from step 6 10. Reboot host with permissive mode(add "enforcing=0" in kernel cmdline) 11. On the host: restorecon -r / 2 failed message: restorecon set context /dev/.initramfs/live->system_u:object_r:device_t:s0 failed:'Read-only file system' restorecon set context /sys/fs/cgroup->system_u:object_r:cgroup_t:s0 failed:'Read-only file system' 12. Activate host in Engine 13. Check that everything is working Additional info: please refer to the attachment audit for /var/log/audit/audit.log
Created attachment 1141191 [details] audit
Just to clarify: Everything is working after step 13?
(In reply to Fabian Deutsch from comment #14) > Just to clarify: Everything is working after step 13? More accurately, after step 13, the prior(on el6) settings on el7 are all working well(permissive mode): 1. Network/ssh work well (the management network is "rhevm") 2. Host is up in Engine Additional info: After step 13, I setup SNMP/CIM/Logging/kdump/RHSM on el7 just now , all have error report, and setup failed.
(In reply to Huijuan Zhao from comment #15) > Additional info: > After step 13, I setup SNMP/CIM/Logging/kdump/RHSM on el7 just now , all > have error report, and setup failed. For these errors - please provide all the logs.
Created attachment 1142068 [details] log
(In reply to Fabian Deutsch from comment #16) > (In reply to Huijuan Zhao from comment #15) > > > Additional info: > > After step 13, I setup SNMP/CIM/Logging/kdump/RHSM on el7 just now , all > > have error report, and setup failed. > > For these errors - please provide all the logs. please refer to the attachment "log" for all the logs, thanks
Huijuan, can we get access to such a host?
This request has been proposed for two releases. This is invalid flag usage. The ovirt-future release flag has been cleared. If you wish to change the release flag, you must clear one release flag and then set the other release flag to ?.
Guys, do we still care about this bug? since 3.6 does not really matter today. And we have this for RHVH 4.x migration: https://bugzilla.redhat.com/show_bug.cgi?id=1421098
(In reply to Marina from comment #24) > Guys, do we still care about this bug? since 3.6 does not really matter > today. And we have this for RHVH 4.x migration: > https://bugzilla.redhat.com/show_bug.cgi?id=1421098 Sandro, could you please help to ack this question? Thanks!
Agreed, marking this as duplicate of bug #1421098 since the migration to el7 will happen with NGN 3.6 *** This bug has been marked as a duplicate of bug 1421098 ***