My next odd ball question(s)....
The SCSI protocal allows for multiple (at least two) controllers on a chain. Normally you would have a controller on one end and a terminator on the other. In this case you have controllers on both ends, and the drives in the middle.
I've seen a friend do this set up with a keyboard(synthisizer) on one side and a windows box on the other, and it seemed to work well.
I would like to do something like this, but with a couple linux boxes sharing the drives. The idea is to create a pair od redundant servers. To start with I would have two high capacity drives on the chain running raid-1. Both systems would have unique IP's but then one would have an aliased addres. If the first system goes down, the second would sense it (ping test?) and alias the address to it's on network connection. If one linux system goes down, the other can assume the roll of the first and as they would share drives all the data would be up todate.
The selling points are, it would be faster than NFS, and would take less rack space in the colo.
Have any of you tried this under linux? Was it stable? If one system gets turned off, will the SCSI chain still be terminated?
The SCSI protocal allows for multiple (at least two) controllers on a chain. Normally you would have a controller on one end and a terminator on the other. In this case you have controllers on both ends, and the drives in the middle.
I've seen a friend do this set up with a keyboard(synthisizer) on one side and a windows box on the other, and it seemed to work well.
I would like to do something like this, but with a couple linux boxes sharing the drives. The idea is to create a pair od redundant servers. To start with I would have two high capacity drives on the chain running raid-1. Both systems would have unique IP's but then one would have an aliased addres. If the first system goes down, the second would sense it (ping test?) and alias the address to it's on network connection. If one linux system goes down, the other can assume the roll of the first and as they would share drives all the data would be up todate.
The selling points are, it would be faster than NFS, and would take less rack space in the colo.
Have any of you tried this under linux? Was it stable? If one system gets turned off, will the SCSI chain still be terminated?
