Yesterday afternoon I all of a sudden started to get GSODs (Grey Screen Of Death) – the equivalent of the Windows Blue Screen of Death. Not good – actually those were my first serious crashes on this particular machine ever (which is about 2.5 years old).
I went through the usual diagnostics, OSX permissions and repair, fsck from single-user mode etc. but nothing helped. One thing I noticed though was that as soon as I booted back into my Macbook Pro and logged in as my normal user, Spotlight (mds/mdworker) went crazy re-indexing. While it was doing that it was pretty much impossible to use other programs because they randomly crashed. I switched Spotlight indexing of (via mdutil from the shell of your choice) and the system was working much better. Hmm, could it maybe be a RAM issue under higher load? This idea made me dig through the logs and I found quite a lot of segmentation faults in system.log – lots on them from mds, but also from the other programs I was trying to use.
So – a flaky RAM could well be the cause of this – System Profiler was still showing 8 GB (2 x 4 GB) though. Luckily I remembered the good old Apple Hardware Test (start your computer and hold “d”). That’ll boot you into a diagnostics tool that can either do a quick (3-4 mins) or longer (10-15 mins) hardware test of your machine.
And it returned… “4MEM/9/40000000: 0xa9d80d98” – not 100% sure what it meant at that stage, I guessed that MEM would indicate a memory issue – which was confirmed by our good friend Google. So, tried the same with just one of the RAM modules installed and it was easy to enough to find the one that was causing the trouble. Now I’m down to 4 GB (ughs) with a new 4 GB module on order – and the problems are all gone for now.
Also – when I was running the AHT with just the one flaky module inserted, the result was “4MEM/4/40000000: 0xa9d81d98”. I therefore assume that the /4/ indicates something like the position or the bank the module is in, the hex value at the end seems to be a memory address.