[Linux-cluster] Using GFS and DLM without RHCS

Danny Wall Danny.Wall at health-first.org
Mon Mar 31 20:39:33 UTC 2008


I was wondering if it is possible to run GFS on several machines with a
shared GFS LUN, but not use full clustering like RHCS. From the FAQs:


Can I setup GFS on a single node and then add additional nodes later? 
        
        Yes you can. For the initial single node setup, simply setup GFS
        using the nolock locking method. Make sure you create the file
        system with enough journals to support the number of nodes you
        wish to add later. (If you do not add enough, you can add
        journals later, but you must add additional space to the volume
        GFS is on to do so.) 
        
        Once you want to add more nodes, you need to setup the cluster
        infrastructure just as you would in an initial multi-node
        configuration. You also need to modify the gfs superblock with
        gfs_tool to switch it to a multi-node locking setup. Use the
        values you would have given to gfs_mkfs - instead of the '-p '
        flag to mkfs, use 'gfs_tool sb  proto ', and instead of the '-t
        ' flag to mkfs, use 'gfs_tool sb  table '. 
        
        Once these changes and additions are made, fire up the cluster
        infrastructure and mount GFS. 
        

I would assume the answer is no, but since this page was published in
2004, I was hoping it is now possible. I would prefer to have a Cisco
CSS front the servers and send clients to the preferred avaiable server
for SAMBA shares, as long as the service is available on that server. If
not, it could re-direct to a different server that is available. This
would simplify the servers by not requiring clustering, and they would
only require GFS and DLM for locking. Ideally, when SAMBA 4 is released
with the ability to load balance the workload, I could allow the Cisco
CSS to do full load balancing. Until then, it would simply act like a
DNS change by talking to one server or the other.

I have had a few problems with RHCS, and while it has done its job most
of the time, if I can simplify the set up by simply moving an IP, it
would be easier to manage and potentially more reliable. Fencing could
be available, but if only one server is used at a time, would it be
needed? The only other access to the disk I can think of, would be for
backups reading from another node. Any suggestions would be helpful.

Thanks
Danny Wall



#####################################
This message is for the named person's use only.  It may 
contain private, proprietary, or legally privileged information.  
No privilege is waived or lost by any mistransmission.  If you 
receive this message in error, please immediately delete it and 
all copies of it from your system, destroy any hard copies of it, 
and notify the sender.  You must not, directly or indirectly, use, 
disclose, distribute, print, or copy any part of this message if you 
are not the intended recipient.  Health First reserves the right to 
monitor all e-mail communications through its networks.  Any views 
or opinions expressed in this message are solely those of the 
individual sender, except (1) where the message states such views 
or opinions are on behalf of a particular entity;  and (2) the sender 
is authorized by the entity to give such views or opinions.
#####################################





More information about the Linux-cluster mailing list