I am at a point where my mds can effectively failover upon a hard or
soft reboot. However, the big problem is with getting the OSSs and
clients to mount to a failover MDS. The syntax according to my lustre
manual is something to the effect:
OSS:
mkfs.lustre -ost -fsname=lustrefs -mgsnid=mds[1-2] /dev/sdb1
mount -t lustre /dev/sdb1 /mnt/lustre
Where there is mds failover pair of mds1, mds2 with a lustre filesystem
name of ''lustrefs''. Heartbeat is correctly running on mds1 and
mds2 and
failover of the lustre mds does work.
Clients:
mount -t lustre mds[1-2]:/lustrefs /mnt/lustre
However, this syntax for the mkfs commands does not work and there a
parsing error with the mgsnids (at formatting time):
mkfs.lustre: Can''t parse NID ''mds[1''
mkfs.lustre: exiting with 1 (Operation not permitted)
I have tried replacing ''[]'' with ''{}'' and
this allows for mounting of
the OSSs, but not the clients. Also a '','' between the mgsnids
does not
work in lieu of -mgsnid=mds[1-2].
Thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.lustre.org/pipermail/lustre-discuss/attachments/20080222/b856197b/attachment-0002.html