I used to know this but it slipped from my consciousness. Thought I'd just stick a reminder out there since I could have used one myself:
If your system is having stability problems, notably if they're not entirely consistent, seem to be related to load, or just plain weird, try swapping a known-good power supply.
I've been dealing with kernel oopses and hangs for a while on one server; fairly consistent in manifestation, seemingly somewhat related to network load. Did the usual things including updating the rather old first-rev original BIOS, reinstalling the OS from scratch to a new boot disk, swapping memory DIMMs, etc. But of course, it was an aging power supply. What finally slapped me awake it started to happen on heavy multi-drive disk activity as well. But you don't always get that obvious a clue.
So as the conventional wisdom goes "power supply problems are notoriously difficult to diagnose, because they often look like something else". Check it first before you do something unnecessarily disruptive and possibly not easily reversible. In particular, I think it's a mistake to try a BIOS update before trying this.
If your system is having stability problems, notably if they're not entirely consistent, seem to be related to load, or just plain weird, try swapping a known-good power supply.
I've been dealing with kernel oopses and hangs for a while on one server; fairly consistent in manifestation, seemingly somewhat related to network load. Did the usual things including updating the rather old first-rev original BIOS, reinstalling the OS from scratch to a new boot disk, swapping memory DIMMs, etc. But of course, it was an aging power supply. What finally slapped me awake it started to happen on heavy multi-drive disk activity as well. But you don't always get that obvious a clue.
So as the conventional wisdom goes "power supply problems are notoriously difficult to diagnose, because they often look like something else". Check it first before you do something unnecessarily disruptive and possibly not easily reversible. In particular, I think it's a mistake to try a BIOS update before trying this.
Last edited:
