We had a similar problem at Hetzner, we were using their consumer grade systems and this was over 6 years ago. To Hetzner's credit though, after we decided it had to be a hardware problem (dmesg would log something about cpu states before every crash) we sent in a ticket and within a reasonable timeframe (I think within a day or a couple, it was long ago so don't know for sure) a Hetzner sysadmin went into the BIOS and changed some feature (I suppose he disabled some sleep mode) and the systems ran perfect after that.
I chalked it up to our decision to run our databases on consumer grade hardware and didn't give it a second thought.
I chalked it up to our decision to run our databases on consumer grade hardware and didn't give it a second thought.