Error during distro-sync.
Closed this issue · 4 comments
Trying to convert Centos 8 to Rocky with latest script.
Server is default Centos 8 with following extra packages:
- @idm:DL1
- freeipa-server
- kexec
- htop (epel)
- fish (epel)
Logs:
Running scriptlet: chrony-3.5-2.el8.x86_64 699/1508
Downgrading : ipa-client-4.9.2-4.module+el8.4.0+589+9650b94f 700/1508
Running scriptlet: ipa-client-4.9.2-4.module+el8.4.0+589+9650b94f 700/1508
Running scriptlet: ipa-server-4.9.2-4.module+el8.4.0+589+9650b94f 701/1508
Downgrading : ipa-server-4.9.2-4.module+el8.4.0+589+9650b94f 701/1508
Running scriptlet: ipa-server-4.9.2-4.module+el8.4.0+589+9650b94f 701/1508
Reinstalling : dracut-network-049-135.git20210121.el8.x86_64 702/1508
Downgrading : python3-wheel-wheel-1:0.31.1-2.module+el8.4.0+ 703/1508
Downgrading : python3-virtualenv-15.1.0-19.module+el8.4.0+59 704/1508
Running scriptlet: python3-virtualenv-15.1.0-19.module+el8.4.0+59 704/1508
Reinstalling : kexec-tools-2.0.20-46.el8_4.1.x86_64 705/1508
Running scriptlet: kexec-tools-2.0.20-46.el8_4.1.x86_64 705/1508migrate2rocky.sh: line 646: 2453678 Killed dnf -y distro-sync
Error during distro-sync.
An error occurred while we were attempting to convert your system to Rocky Linux. Your system may be unstable. Script will now exit to prevent possible damage.
I was forced to rollback, so this server is no-longer available for further debugging. If needed I can try again the migration on copy of this server.
It would help to have the full log from /var/log/migrate2rocky.log
From what you've provided I would say that the distro-sync got manually killed, possibly by some watchdog process or something? Are you running any special security software that might be interfering? Maybe it was an OOM issue?
For sure not OOM, checked it. Full log is no longer available. Currently in progress of rollback.
After rollback finished, I could create clone and try reproduce this there. I will provide further details in a hour.
There isn't special software running on this server outside of listed software in my first post.
Well, the snippet you provided explicitly says, "Killed" which references SIGKILL or signal # 9 which cannot be trapped and certainly is not coming from migrate2rocky itself. To verify next time you run migrate2rocky, can you change line 798 (the distro-sync line) to the following?
dnf -y distro-sync || exit_message "Error during distro-sync: $?"
That will confirm if it's being killed by a signal and which signal (but I suspect it's 9).
Unable to find time to reproduce. Will re-open if I see it necessary.