VE 5.0 Issues Noticed

[Ashley]

So have been testing 5.0 in a small cluster to see how thing's are, these are what I have noticed so far.

1/

When trying to do a live migration between two 5.0 Beta servers in a cluster the following error is displayed:

Mar 27 16:58:51 copying disk images
Mar 27 16:58:51 starting VM 184 on remote node 'cn2'
Mar 27 16:58:54 start remote tunnel
Mar 27 16:58:54 starting online/live migration on unix:/run/qemu-server/184.migrate
Mar 27 16:58:54 migrate_set_speed: 8589934592
Mar 27 16:58:54 migrate_set_downtime: 0.1
Mar 27 16:58:54 set migration_caps
Mar 27 16:58:54 set cachesize: 858993459
Mar 27 16:58:54 start migrate command to unix:/run/qemu-server/184.migrate
channel 2: open failed: administratively prohibited: open failed
Mar 27 16:58:56 migration status error: failed

Offline Migration works fine

2/

I have noticed some issues when using the CPU models (host / host pass through) within KVM, when doing so I receive multiple errors from a kernel panic to a Ubuntu VM being stuck at random points during the boot process.

All occasions changing the CPU back to kvm64 resolves the issues instantly.

3/

Using CEPH + KRBD if I have the discard option ticket this causes the VM to run very slow and at some points for the boot process to time out, again un-ticking discard option causes the VM to run as per normal (this is the same with a clean new empty VM running either Linux or Windows)

I am happy to log bug tickets if these are defiantly "bug's" to me seem mostly new kernel version related.

 

[fabian]

1 is a known bug in Debian Stretch's ssh (as a workaround if you have a private network / only run tests, you can use the plain TCP migration)

feel free to file bug reports for 2 and 3, please include "pveversion -v" output and any relevant config files and logs! thanks

 

[Ashley]

Done as a note.

2) https://bugzilla.proxmox.com/show_bug.cgi?id=1329

3) https://bugzilla.proxmox.com/show_bug.cgi?id=1330