nfsserver resource agent bind umount /var/lib/nfs for nfs_shared_infodir about resource-agents

[higashik]

Hi,

I had the same problem.

I think , bottom causes of this problem are below.

1. Keyword for grep in unbind_tree() , "rpc_pipefs on /var/lib/nfs/rpc_pipefs" has not worked.
   Because mount command displays rpc_pipefs like below.

   sunrpc on /var/lib/nfs/rpc_pipefs type rpc_pipefs (rw)
   ^^^^^^

2. rpcidmapd service will start automatically by nfsd init script (/etc/init.d/nfs).
   But nfsd init script does not stop the rpcidmapd service(with RHEL5/6).
   In addition, rpcidmapd service use /var/lib/nfs/rpc_pipefs.
   So , rpcidmapd continue to use /var/lib/nfs/rpc_pipefs , and an error occurs.

I have confirmed these with RHEL5/6.

I am making patch like below for solve cause 1.

@@ -161,8 +183,8 @@ bind_tree ()

 unbind_tree ()
 {
-   if `mount | grep -q "rpc_pipefs on /var/lib/nfs/rpc_pipefs"`; then
-       umount /var/lib/nfs/rpc_pipefs
+   if `mount | grep -q " on $rpcpipefs_umount_dir"`; then
+       umount -t rpc_pipefs $rpcpipefs_umount_dir
    fi
    if is_bound $fp /var/lib/nfs; then
        umount /var/lib/nfs

I was expect device name from Keyword for grep and add type with -t option.
$rpcpipefs_umount_dir is a new variable , default is /var/lib/nfs/rpc_pipefs.

For solve cause 2 , I've added new LSB resource for control the rpcidmapd service.

my crm nfsd and rpcidmapd part :

group grpNfs \
    prmExNfs \
    prmFsNfs1 \
    prmIpNfs \
    prmNfs \
    prmInitRpcidmapd
(snip)
primitive prmNfs ocf:heartbeat:nfsserver \
    params \
        nfs_init_script="/etc/init.d/nfs" \
        nfs_notify_cmd="/usr/sbin/sm-notify" \
        nfs_shared_infodir="/export/nfs" \
        nfs_ip="192.168.100.12" \
    op start interval="0s" timeout="60s" on-fail="restart" \
    op monitor interval="10s" timeout="60s" on-fail="restart" \
    op stop interval="0s" timeout="60s" on-fail="fence"

primitive prmInitRpcidmapd lsb:rpcidmapd \
    meta \
        migration-threshold="2" \
    op start interval="0s" timeout="60s" on-fail="restart" \
    op monitor interval="10s" timeout="60s" on-fail="restart" \
    op stop interval="0s" timeout="60s" on-fail="fence"

In my environment It work well with these.

I think Lazy unmount work well too, but it is better to deal with the root cause.

Thanks,
Kazuhiko HIGASHI

[higashik]

I made a patch of nfsserver RA for solve cause 1.

#123

P.S.
This patch also has another purpose of keeping rpc_pipefs local (e.g. /var/lib/nfs/rpc_pipefs -> /var/lib/rpc_pipefs ).
So I added a new parameter for specify a mount point of the rpc_pipefs.

[dmuhamedagic]

Many thanks for the report and the patch! @Sov1et , can you please try with the new nfsserver RA.

[dmuhamedagic]

@Sov1et did you get around to testing this? Can we close the issue?

[dmuhamedagic]

Looks like this has been fixed.