Migrate EC2 Instance

[JaggedJax]

Not sure if this is a dumb question or not, but would migrate2rocky theoretically work for an AWS EC2 instance that was started from a Centos 8 image, or are there some behind the scenes things that EC2 is doing that would definitely cause issues.

[nazunalika]

It should work. We migrated a large chunk of our CentOS 8 infra to Rocky 8 in AWS and haven't had issues. Wouldn't hurt to take backups/snapshots, though.

[JaggedJax]

I definitely will take backups and snapshots. Thanks for the info.

[JaggedJax]

I made an image first, but the migration seems to have worked great. I just had to manually remount one EFS drive afterwards, but other than that, all tests I have passed.

[pajamian]

That's great to hear! As we're always looking to improve migrate2rocky, can you elaborate on the issue with the EFS drive?

[JaggedJax]

I am migrating another and will give better notes. But after the restart the EFS drive did not mount, although it was in the fstab.

Before trying anything, I reinstalled this tool (https://github.com/aws/efs-utils) and then performed an unmount and then a mount command and it mounted properly. On this next migration (same exact config) I will check for logs and just try remounting right away if I have this issue and see if it works.

[JaggedJax]

Error message after rebooting:

Dec 31 02:11:00 cioapp103 systemd[1]: Started amazon-efs-mount-watchdog.
Dec 31 02:11:00 cioapp103 kernel: FS-Cache: Loaded
Dec 31 02:11:00 cioapp103 kernel: FS-Cache: Netfs 'nfs' registered for caching
Dec 31 02:11:00 cioapp103 kernel: Key type dns_resolver registered
Dec 31 02:11:00 cioapp103 systemd[1]: user_content.mount: Mount process exited, code=exited status=1
Dec 31 02:11:00 cioapp103 systemd[1]: user_content.mount: Failed with result 'exit-code'.
Dec 31 02:11:00 cioapp103 systemd[1]: Failed to mount /user_content.
Dec 31 02:11:00 cioapp103 systemd[1]: Dependency failed for Remote File Systems.
Dec 31 02:11:00 cioapp103 systemd[1]: remote-fs.target: Job remote-fs.target/start failed with result 'dependency'.

I confirmed that the EFS drive won't mount until I manually re-install the efs-utils package. My installation steps are:

git clone https://github.com/aws/efs-utils
sudo dnf install git make rpm-build
cd efs-utils
sudo make rpm
sudo dnf install ./build/amazon-efs-utils*rpm
sudo mount -a

[nazunalika]

I can see why this has happened. 10 days ago, Rocky was officially merged into amazon-efs-utils. I wonder if you update amazon-efs-utils before you migrate if that would avoid the issue altogether.

[pajamian]

If you have another one and can try what Luis suggested then I'll wait on those results and then add a note to the known issues section of the README file for others who come across this.

[JaggedJax]

Yes, that solves the issue. I was performing the efs-utils update only after the migration when running into the issue, but updating efs-utils before the migration works without issue.

Edit: Pretty convenient timing for me!

[pajamian]

Thanks. I'll make a note of it under known issues, both the recommendation to update efs-utils first and the fix of reinstalling them after (probably updating them after will work as well).