From: Martin Glass <Martin_Glass@tn...> - 2003-12-08 17:34:07
Thanks! My two-node iscsi cluster fails over nicely now.
> > All,
> > While I have got openssi running on iscsi, I have not=20
> been able to=20
> > get
> > failover to my secondary node to work yet (I only have two=20
> nodes, and
> > fail the first by powering it off). I get a "Kernel panic: No
> > secondaries for CLMS failover!", after about 10 seconds. I ran
> > cfs_setroot after the second node comes up; my=20
> /etc/clustertab has both
> > nodes listed as "Y" for CLMS master, and I set a boot=20
> device. (I have
> > had to customise initrd's linuxrc to get iscsi working, so=20
> I didn't use
> > the new initrd chnode creates; I didn't see any difference though.)
> There is a difference. Mount-up your modified ramdisk on=20
> /mnt and edit=20
> /mnt/etc/boottab. Here is a sample two node cluster. Notice both=20
> nodes have a boot device and are listed on the first line.
> This file is based on a /etc/clustertab that looks like:
> 1 192.168.0.63 00:04:5A:4E:3C:72 E 1 =20
> 2 192.168.0.64 00:04:5A:4E:F7:BB E 1 =20
> > How can I tell if the cluster knows my second node is a secondary=20
> > CLMS
> > master? Is there anything else I can check for?
> It didn't get the chnode changes. See above.