A Scary Expiration Episode with AWS and Plesk Free

Recently, I encountered a disconcerting message on the “free” Plesk control panel about its impending discontinuation.

Why am I utilizing this?

The Plesk migration had succeeded previously, so I presumed it would work again (see Migrating Plesk To AWS for more on that saga).

Always eager to explore new avenues, I decided to experiment with LightSail. After a few initial missteps, I managed to get a basic WordPress multi-site operational.

I opted to use UpdraftPlus for the migration. After some configuration and plugin installations, I believed I was prepared.

On the inaugural backup, I received peculiar log messages about quotas and other anomalies. A day or two later, I realized my instance disk was nearing capacity, causing the failures.

Having identified the issue, I intended to remove some files, but I encountered an obstacle: the disk was too full to grant access.

I rebooted several times and gained access for brief moments, but never long enough to locate and delete the problematic files.

Extending the EBS disk

I had faced this issue before, so I knew I could augment the Virtual Drive (EBS Volume) by increasing its size in AWS. Once I did that, I had to execute some commands to get Linux to utilize the new space.

I thought I could follow the same steps I’d used previously:

1. Halt the malfunctioning instance.
2. Initiate a new instance (in the same AZ, with an encrypted drive).
3. Detach the volume from the malfunctioning instance.
4. Attach the drive to the new instance.
5. Execute the Linux commands to expand the volume.
6. Halt the new instance.
7. Detach the expanded volume.
8. Reattach it to the original instance.

First snag: Marketplace security

The “free” Plesk had expired in the AWS Marketplace long ago. Running instances can continue to use the image, but new ones can’t be launched.

I didn’t realize this also prevents the new instance from attaching the volume, with a warning about the AMI being unavailable.

So, a bit of a conundrum… I thought I had bricked myself, but since I hadn’t deleted the “bad” instance, I performed a bit of a shuffle:

1. Shut down the new instance with its pristine volume.
2. Detach its boot drive.
3. Attach the new boot drive to the original instance.
4. Attach the full disk to that same instance (which didn’t error because it was for the original Marketplace listing).

With that resolved, I ran the commands to expand the drive, halted the instance, removed the volumes, re-added the expanded drive (the one that was full before), and booted up the instance.

Lesson Learned

This taught me not to disregard update messages and that my disk space monitoring needs significant improvement.

While I now know how to extricate myself from this predicament, I hope I never have to repeat this.

Migration ahead

All that remains is to complete the migration. I still plan on migrating from Plesk to a simple WordPress server, as Plesk is excessive for my use case (and revenue).