[ENBD] RAID

[robert stanford]

Hi everybody, 
well we're about to roll out our backup server running nbd server. Just
have a couple of questions and would like to hear about some of the ways
people have implemented their network raid arrays. 

Anyhow heres what we have. 

The main server is a dual PIII-666 with 2x40Gb IBM drive running raid1,
the main application it runs is filemaker server to about 25 POS
terminals. 



raiddev             /dev/md0 
raid-level                  1 
nr-raid-disks               3 
chunk-size                  64k 
persistent-superblock       1 
nr-spare-disks              0 
    device          /dev/hda3 
    raid-disk     0 
    device          /dev/hdc3 
    raid-disk     1 
    device          /dev/nda
    raid-disk     2

/var = md0  <--this is what we will be mirroring via nbd, /dev/hda3 , 
/dev/hdc3 , /dev/nda will be its 3 mirror devices after I reconstruct
the md. 

I have a test system in the workshop where we have been playing around
and have modified the startup scripts so that the network starts before
the nbd server and then raid is started and /var is mounted. Things
seems to be working fine. 

These is one thing I am worried about though, and i'm sure its just
because i'm still new to working with raid. 

If I unplug the crossover cable the raid device that is relying on the
nbd hangs, of couse I can do raidgenerate error and raidhotremove and
raidhotadd, however the thing that concers me is if the backup machine
hosting the nbd server goes down or the crosover cable is unplugged etc.
, then the main machine hangs on the broken md. Is there supposed to be
any detection of the broken block device ?

I was looking at using some sort of heartbeat between the machines ,
however at this point I'd be realy interested in seeing how some other
people have used enbd's and especialy how they handle the raid arrays
associated with them. 


Regards 
Robert Stanford