I have a two node cluster setup to be a HA nfs server for several zpools. Everything is going fine according to all instructions, except I can not seem to find any documents that discuss nested zfs mounts. Here's what I have:
-----
top-level zfs:
# zfs list t1qc/img_prod
NAME USED AVAIL REFER MOUNTPOINT
t1qc/img_prod 18.8M 5.88T 1.07M /t1qc/img_prod
- descendant from that zfs are many other zfs's (i.e., t1qc/img_prod/0000, t1qc/img_prod/0001.... etc.)
-----
the top-level is setup under the HAStoragePlus as follows:
# clresourcegroup create -p PathPrefix=/t1qc/img_prod improdqc1-rg
# clreslogicalhostname create -g improdqc1-rg -h improdqc1 improdqc1-resource
# clresource create -g improdqc1-rg -t SUNW.HAStoragePlus -p Zpools=t1qc improdqc1-hastp-resource
# clresourcegroup online -M improdqc1-rg
# clresource create -g improdqc1-rg -t SUNW.nfs -p Resource_dependencies=improdqc1-hastp-resource improdqc1-nfs-resource
# clresourcegroup online -M improdqc1-rg
- contents of /t1qc/img_prod/SUNW.nfs/dfstab.improdqc1-nfs-resource:
share -F nfs -o rw -d "t1qc" /t1qc/img_prod
-----
jump over to one of my other servers (linux rhel5) and mount the exported fileystem
# mount -t nfs4 improd1:/t1qc/img_prod /zfs_img_prod
- that works just fine, if I execute an ls of that mounted filesystem, I see the listings for the descendant zfs's, however, if i try to access one of those:
# ls /zfs_img_prod/0000
ls: reading directory /zfs_img_prod/0000: Input/output error
-----
jump over to one of my solaris 10 servers and mount the exported fileystem
# mount -F nfs -o vers=4 improd1:/t1qc/img_prod /zfs_img_prod
- that works just fine, if I execute an ls of that mounted filesystem, I see the listings for the descendant zfs's, however, if i try to access one of those:
# ls /zfs_img_prod/0000
#
- empty listing, even if there are files/directories in that zfs on the server
-----
This setup worked great without the cluster, i.e., just shared with zfs. Is this not possible under the cluster or am I missing something?
Thanks.