Distinguished colleagues,
I have to poll this good forum again. I noticed a couple of posts recently on this subject, so I thought I could stir the pot a little with a question from a slightly different angle.
Can a spectrometer host computer be set up with a RAID configuration and will the integrity of the NMR data be preserved IF a disk does indeed go down during acquisition? With microsecond NMR pulses and delays, I would think that the redundant disk taking over, will have to do so in an extremely quick manner, would it not? Can this be accomplished? I have been to a few ENC's and SMASH's in my time and cannot recall any mention of RAID, well, ever. Am I wrong and is my memory letting me down?
Our O2's have recently rebooted themselves overnight on weekends a few times, and root receives mail referring to the SGI supplied crontab file command 'find ... exec rm...' which has me a little worried; commenting this command out. leaving the disk in question unmounted and doing repeated xfs_check's and xfs_repair's with no errors reported, has eased my worries and no spontaneous self-reboots since.
My IT/IS response, though, has been as follows, partly also in the name of Sarbanes-Oxley(!):
Usually on bussiness critical systems, when I.T. places the order, We plan ahead and make sure the equipment comes with a raid configuration. We shouldn't have to swap out an entire system for a simple disk drive failure, to tell you the truth.... I'm a bit surprised this wasn't explained to you when you got your unix certs!!! This sort of stuff will continue to hapen to you as long as you keep installing systems without IS supervision.)
Sorry, couldn't resist this last (admittedly: sarcastic) jab at my IT support (or whatever you want to call it), I just try to provide the best NMR support I can to my 60-70 chemists here and it ain't always a piece of cake...
Jens Knudsen
Arena Pharmaceuticals
Received on Mon Oct 31 2005 - 11:55:48 MST