At work this week I was trying to get a Linux box to recover from a broken file system. In the process I realised that the dual Pentium Xeon box was running a non-SMP kernel. It looks like work bought 6 dual CPU boxes three years ago, for the SAP system, but never installed a SMP kernel on them. From what I understand, this means that one of the CPUs in each box has been sitting idle for the last three years...
I don't like to critisise, the person who runs the SAP system knows SAP very well, but doesn't know Unix or Linux much at all. However his boss should have arranged better training... I'm only a RHCT, hardly a RHCE or a seasoned Unix veteran, but I spotted the installation flaw.
Re:backups
ajt on 2005-06-13T20:19:32
Though there is backup for most systems, the app servers hold no data, so in this case are not actually backed up. The solution was to stop SAP and change to single user mode, fsck the problem partition, and it was happy again. The hardware RAID was happy, so I suspect SAP did something stupid, and was mostly a case of stopping SAP and cleaning up after the fallout.
After prodding the system a bit more, I have made sure that they are actually running SMP kernels, so I made an error there. I'm the most Linux person on site, but clearly not that experienced...
