kumaran.rajaram at gmail.com
Sat Feb 14 13:27:21 GMT 2009
Yes, you would need minimum of two NICs per node (say eth0 and eth1). Devote
one NIC for ctdb/nodes(e.g. eth0) and configure this IP address to be
internal IP address used by your file-system (GFS/Lustre/GPFS). Use the
other NIC for ctdb/public_address (e.g. eth1 or bond0) which is totally
managed by CTDB (IP failover + failback as node fails/recovers). The public
address should be used for NFS/SMB export.
You need to manually start ctdb on all the nodes. I use parallel shell
(pdsh) to start CTDB in parallel across nodes (use eth0) to launch. Pdsh
internally uses rsh or ssh across nodes.
pdsh -a "/etc/init.d/ctdb start"
On Sat, Feb 14, 2009 at 6:06 AM, Ken Lupo <lupok at saline.k12.mi.us> wrote:
> I am trying to understand the ip addressing with nodes and public
> addresses. Do I need to 2 or more ethernet nics in each ctdb node? What is
> best practice for addressing?
> Also, how do you start up ctdb on all nodes at once? with onnode? or can
> you setup the ctdb service to the clsuter manager?
> Thank you,
> On Sat, Feb 14, 2009 at 7:41 AM, Kums <kumaran.rajaram at gmail.com> wrote:
>> On Fri, Feb 13, 2009 at 9:24 PM, Ken Lupo <lupok at saline.k12.mi.us> wrote:
>>> I am not sure if this is the correct list but I am having some issues
>>> CTDB and configuring it properly. I have gone through all the
>>> many many times and am still not able to get it setup and running
>>> Is there a howto on setting up CTDB in a RedHat GFS cluster out there?
>> What is the exact issue that your seeing? Can you please elaborate a
>> little bit.
> Ken Lupo, Saline Area Schools
> Office: 734.429.8014
> Mobile: 248.881.5681
> idea=`grep -i clue /dev/brain` ; test -z "$idea" && echo "sorry, init 6 in
> progress" || sh ./post-to-web
More information about the samba