XCP-ng 8.0.0 Release Candidate

dave.opc

Today i found out, that my xcp was restarted at 3.51AM. There is no crash dump in xcp center. Where do i look to find out what happened?

r1

@dave-opc whats the XCP-ng version?

Did you see -

https://github.com/xcp-ng/xcp/wiki/Troubleshooting

https://github.com/xcp-ng/xcp/wiki/Logfiles

dave.opc

@r1
I thought it's obvious if i am asking in 8.0RC1 thread, that i have this verison.

r1

@dave-opc Making sure that the version is up to date.

Did you find anything in /var/log/kern.log?

r1

@EliasSeccom Does ALT + Arrow keys (Left/Right) show you any erroneous messages? On one of the screen, there is a shell available.

If its a hang which is not letting you do anything, you can attach a serial console and see if it outputs any stacktraces.

dave.opc

kern.log.txt

Kern log file shows that it booted at that time, at least i don't see any errors. (file attached)
It looks like it was just reset and started booting, but i doubt there was an electricity outage, as my xcp is powered by ippon back basic 1050.

r1

@dave-opc To know more on next incident, you may add noreboot options to multiboot2 /boot/xen.gz in /etc/grub.cfg file against default kernel boot entry.

dave.opc

Got another issue with RC1:

I got 2 xcp ng servers. 1st at home on 8.0rc1 and 2nd remote on 7.6
When i had at home 7.6 (before upgrade) my continuous replication task in XOA was working fine. Every night copying VM to home server.
After upgrading to 8.0RC1 XOA starts copying. I can see [importing....] in xcp center, but after transfering the file, xoa ends with error
I thought i had a problem with iscsi drive, connected to xcp, but i also tried copying to my main sr (nvme) and got the same error.

olivierlambert

You are doing CR from 8.0 to 7.6?

dave.opc

@olivierlambert
from 7.6 to 8
P.S. just checked. the other way from 8 to 7.6 CR is working fine.
Also copying VM from 7.6 to 8 is working. - problem with CR 7.6 -> 8

olivierlambert

Okay so it's maybe something related to how VDI are handled in the "opposite" way (newer to older).

I would suggest to get everything at the same level if possible. Ideally, try to test from a 8.0 to a 8.0 to be sure it's a version diff that cause the issue.

dave.opc

@olivierlambert
Of course i will try that, but can't install 8.0 on production pc yet. Will wait for 8.0 release.
I just posted this issue, so maybe you or someone will try to do CR 7.6 -> 8.0RC1 and see if the issue still exists.

stormi

XCP-ng 8.0 will now have the EPEL repositories added - but disabled - by default, in order to avoid users add it and not notice that adding it automatically leaves it enabled.

I have also written this detailed page that attempts to cover the subjet of installing additional packages to dom0: https://github.com/xcp-ng/xcp/wiki/Additional-packages

By the way, various new utilities have been added to XCP-ng repositories for 8.0, I won't list them here, but you can find the answer in that page (how to lure people into reading it...)

axel

Hello

now it looks like 8.0 will be ready soon. I like to write a article for a big german IT magazin

stormi

We plan to release in the next few day, and there have been important changes that need testing before that:

the updater XAPI plugin, that handles patch detection and application (through yum) and is used by Xen Orchestra in the pool and hosts views, has been vastly modified to prevent it from running multiple times concurrently (in extreme cases, this could crash your host due to lack of available RAM)
updated kernel for http://xenbits.xen.org/xsa/advisory-300.html
EPEL repos added by default, in disabled state http://xenbits.xen.org/xsa/advisory-300.html

Testers who already run XCP-ng 8.0 beta/RC

To test this, I suggest to:

Update the XAPI plugins only, so that you can test the updater plugin: yum update xcp-ng-xapi-plugins
Update your hosts with Xen Orchestra (if you don't use Xen Orchestra, then you won't be able to test the updater plugin... then just run yum update on your hosts).
Restart the XAPI toolstack (xe-toolstack restart) on your hosts, starting with the pool master.
Reboot your hosts, starting with the pool master as usual, so that the kernel security fix is applied and you can test that everything works well with the updated kernel.

Testers who could test the "final" installation ISO

I'll also make pre-release ISOs available to those who want to test them before the final release (we need as many tests as we can get in this final stage). Unless there's an important bug that is found during the tests, the pre-release ISOs and the final ISOs should be identical.

stormi

The "prefinal" ISO images are available there:

The net-install online repository has been updated too.

axel

Hello stormi

Thx for your info i will try your pre release next few days

ajmind

Upgrade from XCP-NG 7.6 > 8.0.0 with prefinal .iso went smooth without any problem on a single host. Host system quite old Intel System S5520UR with dual XEON E5645 and 48GB RAM.

Just a quick note: Old Windows XP VM does not boot anymore with XenTools 6.5 or 7.3 installed. When removed in safe mode the VM can be started... On XCP-NG 7.6 this was not the case. VM has booted also with XenTools 6.5.

janvs0

Upgrade from Citrix XenServer 6.5 > Citrix XenServer 7.6 > XCP-ng 8.0.0 with prefinal .iso went smooth without any problem on a 3 box pool. Moved to the "new" partition scheme from XS6.5 -> XS7.6 and used the prepare_host_upgrade.py (rolling pool) method. Had no issues migrating the existing VM's upward.

Perhaps it would've been possible to go from XS6.5 straight to XCP-ng 8.0.0 (after applying hotfix RPU002) but since this is a production cluster I didn't want to make it too risky (yes, I know this is not quite final yet).

//edit: Forgot to mention these are Dell PowerEdge R720s on Intel E5 2660's using FC storage.

Thanks XCP-ng team!

mike

Quick (hopefully not dumb) question: Will XCP-ng 8.0 support VDIs larger than 2TB? I see mention of the qcow2 format dotted around but nothing concrete about if/when it will be implemented. Cheers.