Migration to Podman: Fatal errors
digdilem opened this issue · 3 comments
Problem description
After spending a day, I've now abandoned trying to migrate Uyuni to Podman on Rocky and am now following the "tested" method of using Leap Micro 5.5
Things definitely went smoother, but still errored out without completing, leaving a broken new host.
Process: New vm. 150Gb dis, 16G ram, 4x cpus. Installed Leap-Micro-5.5
Follow this guide: https://www.uyuni-project.org/uyuni-docs/en/uyuni/installation-and-upgrade/container-deployment/uyuni/migrate-uyuni-to-a-container.html
Setup ssh etc, all good.
Run prepare: mgradm migrate podman --prepare --logLevel debug oldserver.fqdn
This completes without errors first go. Feeling confident.
It tells me to run again without --prepare, so I do so.
It then exits after ten minutes or so with:
Error 1:
Error: cannot generate systemd service file: failed to generate systemd service unit file: /etc/systemd/system/uyuni-server.service file already present, not overwriting
I move this file out of the way and re-run.
Error 2
The script then stops here.
Checking cluster versions ok
pg_controldata: fatal: could not open file "/var/lib/pgsql/data-pg14/global/pg_control" for reading: No such file or directory
could not get control data using "/usr/lib/postgresql14/bin/pg_controldata" "/var/lib/pgsql/data-pg14": child process exited with exit code 1
Failure, exiting
Error: cannot run PostgreSQL version upgrade script: failed to run uyuni-upgrade-pgsql container: exit status 1
Has this migration script been tested recently? It seems very fragile.
Steps to reproduce
As above
Uyuni version
Uyuni 2024-08
Uyuni proxy version (if used)
No response
Useful logs
No response
Additional information
No response
Update:
I removed the various systemd files and all of /var/lib/container/storage on the new host and then re-ran the migration command WITHOUT --prepare.
This time it ran through without error.
So it looks like running --prepare before migration causes the errors in this ticket.
Closing as I've self-resolved the issue
The first error should be fixed in master
. I can't recall having seen the other one...