Ok I was not quite sure where to put this so forgive me if I am posting in the wrong forum. First a little detail as to what I am trying to do.
I have built a 7 drive ( soon to be eight ) FC-AL Jbod with 4 9.1 gig 10k drives, 2 36.7 gig 10k 73lp series drives, and one 36.7 10k drive with 16 meg of cache. All were bought relatively cheap with the 9.1 gig drives costing a mere 15 bucks a piece, and the other three in the 120 dollar neighborhood. the neccesary t-cards were bought from HEREThe hard drives are in the following config, the 4 9 giggers are software striped together, the two lps again are software striped, and the 36.7 gig drive exists as a single basic disk. Together this setup is connected via a qlogic 2100 series hba ( $50.00 ). All sits and exists just find and dandy with some pretty impressive benchmarks on one machine. Now here is were it gets tricky. Fibre channel is a dual arbirtrated loop. This means I can run off the "b" channel the same group of disks to another machine which I have done. Now the drives are local drives on each machine. So far so good, everything aok. The config is below:
Machine 1
drive c 20 gig IDE (OS)
drive d 36 gig FC-AL ( 4x9.1) RAID O
drive e 72 gig FC-AL (36.7 x 2) RAID O
drive f 36.7 gig FC-AL ( OS of machine 2)
Machine 2
drive C 36.7 gig FC-AL (OS) same as drive f on machine 1
drive D 72 gig FC-AL Raid O same as drive e on machine 1
( i have not yet put the 4 drive array on loop b yet)
Here is the issue. When machine 1 writes to one of the shared drives (e) the data cannot be seen by machine 2 until a reboot. The same can be said in reverse. After writing data on one machine, a reboot on the other will cause a checkdisk, sometimes making the drives corrupt, sometimes finding the data written by the other machine and continueing normally. Data can be read from the shared drives with no issues what-so-ever. Benchmarks as well can be run for either machine with no ill effects. I know this to be an indexing issue with 2k xp, but really don't know how to set this up to work properly any ideas would be much appreciated.
I have built a 7 drive ( soon to be eight ) FC-AL Jbod with 4 9.1 gig 10k drives, 2 36.7 gig 10k 73lp series drives, and one 36.7 10k drive with 16 meg of cache. All were bought relatively cheap with the 9.1 gig drives costing a mere 15 bucks a piece, and the other three in the 120 dollar neighborhood. the neccesary t-cards were bought from HEREThe hard drives are in the following config, the 4 9 giggers are software striped together, the two lps again are software striped, and the 36.7 gig drive exists as a single basic disk. Together this setup is connected via a qlogic 2100 series hba ( $50.00 ). All sits and exists just find and dandy with some pretty impressive benchmarks on one machine. Now here is were it gets tricky. Fibre channel is a dual arbirtrated loop. This means I can run off the "b" channel the same group of disks to another machine which I have done. Now the drives are local drives on each machine. So far so good, everything aok. The config is below:
Machine 1
drive c 20 gig IDE (OS)
drive d 36 gig FC-AL ( 4x9.1) RAID O
drive e 72 gig FC-AL (36.7 x 2) RAID O
drive f 36.7 gig FC-AL ( OS of machine 2)
Machine 2
drive C 36.7 gig FC-AL (OS) same as drive f on machine 1
drive D 72 gig FC-AL Raid O same as drive e on machine 1
( i have not yet put the 4 drive array on loop b yet)
Here is the issue. When machine 1 writes to one of the shared drives (e) the data cannot be seen by machine 2 until a reboot. The same can be said in reverse. After writing data on one machine, a reboot on the other will cause a checkdisk, sometimes making the drives corrupt, sometimes finding the data written by the other machine and continueing normally. Data can be read from the shared drives with no issues what-so-ever. Benchmarks as well can be run for either machine with no ill effects. I know this to be an indexing issue with 2k xp, but really don't know how to set this up to work properly any ideas would be much appreciated.
