[Linux-cluster] Simplest 4 node GFS 6.1 cluster

Troy Dawson dawson at fnal.gov
Fri Jul 8 17:44:56 UTC 2005


Sorry for the delay, I was on vacation, then catching up from being on 
vacation.

Basically, our Fiber Channel switch is hands off to me.  Not that I 
don't have access to the password, but there are several much more 
important things connected to that switch.  If I were to mess them up, 
or cause them to go down for a bit ... well that wouldn't be good.

It looks like I'm going to be going with the power switch option.  The 
rack my macines are in doesn't currently have controllable power, put 
it's in the works, so I guess I'll just be patient.

Troy

JACOB_LIBERMAN at Dell.com wrote:
> Troy,
> 
> What is the model/vendor of the switch? What are the models/vendors of
> the servers? I ran into a similar problem in our lab (no network power
> switches) and there are a few ways to get around that problem.
> 
> Thanks, jacob 
> 
> 
>>-----Original Message-----
>>From: linux-cluster-bounces at redhat.com 
>>[mailto:linux-cluster-bounces at redhat.com] On Behalf Of Troy Dawson
>>Sent: Thursday, June 23, 2005 8:43 AM
>>To: linux clustering
>>Subject: [Linux-cluster] Simplest 4 node GFS 6.1 cluster
>>
>>Hello,
>>I've really been trying to figure this out from the 
>>documentation, and if it's in there, please point me at it.
>>
>>Here is what I want to do.  I have 4 machines, all connected 
>>to the same SAN disk.  1 machine will be the only machine 
>>that can read and write to the disk.  The other 3 will be 
>>read only.  The read only machines will be the only ones that 
>>have outside services running, such as ftp, nfs, rsync.
>>
>>I don't have a reliable 5th machine.  So I don't have a way 
>>to do an external lock manager.
>>
>>I don't want any failover, but I also don't want a single 
>>point of failure.  Basically if one of the read only machines 
>>goes dead, then it's dead, nothing takes it's place, but the 
>>other machines can go right on working.  If the read, write 
>>machine goes dead, then it's dead, but the read only machines 
>>can go right on doing what they normally do.  And if two or 
>>three machines die, I still want the one to still be able to 
>>at least read the data.
>>
>>This idea of fencing is what's throwing me off.  If I'm 
>>reading things right, I can't do group GFS without them being 
>>in a cluster, and they can't be in a cluster without doing 
>>fencing.  But the fencing seems to just allow the various 
>>machines to take over for one another.
>>
>>I also don't have access to the SAN switch, other than my 
>>machines plug into it.  It's essentially a black box.  These 
>>machines also don't have any way to remotely turn power on an off.
>>
>>Is GFS what I really want?  I've tried just standard ext3, 
>>but I was getting a caching problem with my read only 
>>machines.  Do I just want to try and fix my caching problem?
>>
>>Troy Dawson
>>--

-- 
__________________________________________________
Troy Dawson  dawson at fnal.gov  (630)840-6468
Fermilab  ComputingDivision/CSS  CSI Group
__________________________________________________




More information about the Linux-cluster mailing list