I might have temporarily solved the issue
The issue of the crashing server wasn’t an as easy fix as I thought. The crashing database table wasn’t the culprit at all. The CPU kept hanging at irregular intervals anyway. I was running on two CPU cores and decided, just because I ran out of options, to downgrade it to run on just one CPU core. Speed wise we’re not missing out on all that much currently – it’s still very snappy and the system load is at 0.42 when the old system was going between 10 and 20 the last weeks. Much to my surprise the system has now been up for 17 hours without an issue. Not really ruling out that it might go down at any moment, but it does seem to be behaving now.
If it stays up during the weekend I’ll send over the error messages from the kernel when we ran two CPU cores to the hosting company and see if they have come across it before. It’s an issue that needs to be fixed as I’m currently running the biggest package for 1 CPU they’ve got so if I want to crank it up I have to go for at least two CPU cores.
I will now try to get all the member functions going so that we can start reporting all the results. It’s a busy motorsports weekend as always!