XCP-ng 8.2.1 (maintenance update) - final testing sprint

stormi

The error message suggests an internal network error message. Does this host have custom iptables rules, or xapi.conf entries?

apz

@stormi I have not touched xapi.conf, iptables has one additional rule to default, which allows port 4949 in. It's copy-paste of other rules in /etc/sysconfig/iptables, save changed port.

stormi

@apz Are there differences between the installed hosts that could be a lead? Could you compare the lists of installed RPMs between your hosts? (rpm -qa | sort). Is the network configuration different?

Are there error logs in xensource.log, user.log, daemon.log when you run /usr/bin/create-guest-templates-wrapper?

apz

@stormi I diffed this host (originally installed as 8.2.0) and one I installed very recently (during 8.2.1 testing phase). Their differences are:

137d136
< igc-module-4.20.17-1.xcpng8.2.x86_64
274a274
> MegaCli-8.07.14-1.noarch
394d393
< r8125-module-9.003.05-1.xcpng8.2.x86_64

The 8.2.1 installation appears to have network card modules in packages, the old installation has MegaCLI installed because of its RAID controller.

xensource.log is attached, user.log didn't yield anything, neither did daemon.log

As I was running this command again and again, it succeeded once. Out of curiosity, I tried running it again and again on my other hosts that had succeeded the last time. I could introduce a random failure on all of them, maybe 1 out of 20 or more rare.

xensource.txt

stormi

@apz It's not related to your issue, but you need to read https://xcp-ng.org/forum/post/47128

apz

@stormi The other (working) machine is indeed installed with test5.

Should I care about the fact that the testing repo no longer exists, so the two first commands fail?

stormi

@apz If you removed the repo file for staging you can just remove the --enablerepo part.

mortenchristensn

Tried to update my homelab from 8.2 to 8.2.1 with the iso on usb.

After having answered all questions, the installer stops with

Error occurred
An unrecoverable error has occurred. The error was:
Installer mode (UEFI) is mismatched with target boot
mode (legacy)
Reboot

Can I choose between legacy and uefi somewhere?
Do I have to change my bios-settings from legacy to uefi, and can it damage my working installation?

stormi

@mortenchristensn I see two mistakes here:

the process to update from 8.2 to 8.2.1 is not by using the ISO. Check the blog post.
when you want to upgrade a host using the ISO, you need to make sure you boot the ISO in the same mode as the installed system (BIOS vs UEFI).

mortenchristensn

@stormi said in XCP-ng 8.2.1 (maintenance update) - final testing sprint:

@mortenchristensn I see two mistakes here:

the process to update from 8.2 to 8.2.1 is not by using the ISO. Check the blog post.

Can it be rolled back to a working 8.2, if it is updated with "yum update" ?
I have a server with similar motherboard running 8.1. Is that recommended to upgrade with "yum update" too ?

stormi

@mortenchristensn said in XCP-ng 8.2.1 (maintenance update) - final testing sprint:

@stormi said in XCP-ng 8.2.1 (maintenance update) - final testing sprint:

@mortenchristensn I see two mistakes here:

the process to update from 8.2 to 8.2.1 is not by using the ISO. Check the blog post.

Can it be rolled back to a working 8.2, if it is updated with "yum update" ?

Not exactly, but you can downgrade RPMs individually if really needed.

I have a server with similar motherboard running 8.1. Is that recommended to upgrade with "yum update" too ?

In this situation it's your choice. If you want to have a backup you can rollback to, you should favour the ISO.

KPS

Hi!

I am having the same issue, as described in the first "XCP-ng 8.2.1 thread":

fcoe_driver CRITICAL: ['Traceback (most recent call last):\n', '  File "/opt/xensource/libexec/fcoe_driver", line 34, in execute\n    output = subprocess.check_output(cmd)\n', '  File "/usr/lib64/python2.7/subprocess.py", line 575, in check_output\n    raise CalledProcessError(retcode, cmd, output=output)\n', "CalledProcessError: Command '['fcoeadm', '-i']' returned non-zero exit status 2\n"]

8.2.1 is installed and fully patched.
Do you have any idea on how to fix this?

Best wishes

stormi

@KPS Hi. This thread is now inactive sop please open a new thread dedicated to your issue.