Debian update corrupted MariaDB database on Amazon Lightsail

A recent Debian update corrupted 2 MariaDB databases on Amazon Lightsail servers. I run SolydK as my desktop operating system. SolydK is based on Debian Linux and KDE. I recently moved all my Amazon Lightsail WordPress servers to Debian. Once setup – things ran smoothly – until an update last night.

When SolydK alerts me to an update – after updating – I head over to Amazon Lightsail to see if my Debian servers need updated. In some case they don’t. But last night they did. So I ran an update and upgraded. All seem to go well until I rebooted servers as instructed by upgrade process.

In past when using cPanel I always rebooted the VPS ( Virtual Private Servers ) – never Apache etc. This is what I did with Amazon Lightsail – rebooted both servers. And this is when trouble showed up. Dreaded “Could not create a database” error was on full display. I tried a couple more reboots – no luck.

My next thought was to try and reboot the MariaDB database. This offered no solution. Digging a bit deeper – I found that this database had obviously become corrupted. I can only assume something in update, update process or reboot caused problem.

So I was stuck removing and then reinstalling MariaDB. For those who aren’t aware – this means I basically had to rebuild this WordPress site from backups last night / this morning. I had to actually retrieve my latest article from Google cache – because I hadn’t done an XML export of WordPress since writing that article.

Problem I ran into seems to be very common. A quick search engine inquiry shows plenty of others who saw same exact error. Unfortunately none of the many topic threads with supposed solutions worked. So it was easier to just recreate database and restore from backups.

That created a whole bunch of frustrations I had never ran into. When importing XML WordPress backup – server kept timing out. In 15 years of running a WordPress website. I don’t recall ever having this aggravation. Then Cloudflare started acting up with a 524 error. So it took a few hours before I could get XML backup up and running.

When backup finally implemented. It created multiple duplicate post. Which had to be manually removed and then import process started again.

In Firefox it seems main page for this site isn’t displaying properly. I’ll have to get to fixing that this coming week. MetaSlider post of course are screwed up – as always – because of how this plugin is designed and it offers no proper backup process – even with their PRO offering.

I’ll have to spend this next week troubleshooting, finding and resolving any errors I come across. Restoring MetaSlider takes hours – if they would listen to user feedback this wouldn’t be such an issue.

Unfortunately I will also have to fully rebuild another website I admin for someone else. Although they rarely use it – I keep it running and updated.

All this drama was created because I wanted to keep my servers up to date. Keeping servers and operating systems updated is a good security practice. But what I have experienced in past 24 hours is exactly why people refuse to update these things. In many cases – doing so breaks what is running smoothly and then requires tons of work to fix.

In the future before doing any server updates. I will make an Amazon Lightsail Instance backup. So I can revert to that when and if things get fried.

Web hosting Cloud services Domain registration Webdesign

Cloudflare Registrar most painless domain transfer I’ve ever experienced

Facebook YouTube Twitter are NOT your brand

diaspora* the Facebook alternative no one seems to want to use