sharing scsi disk beetween two freesbd's ...
Bill Reid
wer at cstone.net
Thu May 1 02:56:56 PDT 2003
If your machine fails you are still going to have a dirty disk. NFS
would be better depending on how much data you are talking about.
For the NFS machine:
You may wish to just get two SCSI Raid Controllers and a raid cabinate.
Put faith in your raid, redundant controllers and have another failover
machine incase your nfs server goes down.
Then you could have some redundant machines that can mount the raid. $$$$
Depends how much money you have.
There are multiple linux HA projects that use some interesting ways to
duplicate data and handle failover. http://linux-ha.org/
They have a Network Block Device that is really neat. You can tweak the
writing and conformation of writes a little. There are failover
mechanisms available. FreeBSD does not have a network block device
though as far as I know (wich is not a lot).
Ther was a FreeBSD Olympics website that had a shared SCSI RAID device.
It seemed elegant but I have not been able to find it. I had a drawing
of the layout but no other resources.
So we went with the Network Block Device above. It is cheap and should
let me sleep through pages. (I work with linux freaks but I think this
was the right choice for us).
-=Bill
Oliver Fromme wrote:
>Geoffroy DESVERNAY <dgeo at esm2.imt-mrs.fr> wrote:
> > Of course, but the two machines are going to be redundant servers, ant
> > the disc
> > (an IDE raid array presented as scsi disk) will have to be accessible by
> > the second server,
> > in case of failure of the first one...
>
>For that kind of setup, I would recommend to use something
>like a NetApp Filer. It appears as an NFS server by itself,
>so you can mount it on both of your redundant machines.
>Even both read+write, if you want.
>
>You can even cluster two Filers that mirror each other, so
>you get an additional level of redundancy.
>
> > Isn't there any mechanism that could force the system to flush read
> > cache on each operation ?
>
>No. You could of course build your own fail-over mechanism:
>During normal operation, host1 mounts the SCSI disk and
>host2 mounts it via NFS from host1. If host2 detects a
>failure of host1, it forcibly removes the NFS mount and then
>mounts the SCSI disk itself. You have to be very careful,
>though, since there are a few possible pitfalls.
>
> > I think the design of scsi bus, which allow 2 'SCSI adapters' (=> 2
> > computers) should imply that one disc (or device) can be accessed by two
> > hosts at the same time... for reading of course :)
>
>It might imply that in the _hardware_ level. But you still
>need support at the OS level (e.g. VM cache subsystem),
>which FreeBSD does not provide for its UFS/FFS.
>
>Regards
> Oliver
>
>PS: I'm not afiliated with NetApp, but I've been using
>their products and I'm very satisfied with them.
>
>
>
More information about the freebsd-stable
mailing list