Arch Linux

premier

Member

Registered: 2006-12-07

Posts: 16

NFS4 and Idmapd lookup failed when resuming from suspend

I finally figured out why the ID mapping failed when an NFS client resumed from suspend.  Just posting here in case someone else can use this information.

Problem:
The permissions on my NFS4 mounts were being set to nobody:nobody but only after suspending/unsuspending.  Everything worked before going into suspend.  Restarting services on the client didn't help.  Restarting NFS on the server fixed it but that was not an acceptable solution.

Log snippet below:

Mar 31 19:35:12 ArchKnight rpc.idmapd[3979]: New client: 0
Mar 31 19:35:12 ArchKnight rpc.idmapd[3979]: Opened /var/lib/nfs/rpc_pipefs/nfs/clnt0/idmap
Mar 31 19:35:12 ArchKnight rpc.idmapd[3979]: New client: 1
Mar 31 19:35:12 ArchKnight rpc.idmapd[3979]: nss_getpwnam: name '0' does not map into domain 'reading'
Mar 31 19:35:12 ArchKnight rpc.idmapd[3979]: Client 0: (user) name "0" -> id "99"
Mar 31 19:35:12 ArchKnight rpc.idmapd[3979]: Client 0: (group) name "0" -> id "99"
Mar 31 19:35:22 ArchKnight rpc.idmapd[3979]: nss_getpwnam: name '1500' does not map into domain 'reading'
Mar 31 19:35:22 ArchKnight rpc.idmapd[3979]: Client 0: (user) name "1500" -> id "99"
Mar 31 19:35:22 ArchKnight rpc.idmapd[3979]: Client 0: (group) name "501" -> id "99"
Mar 31 19:35:22 ArchKnight rpc.idmapd[3979]: nss_getpwnam: name '1503' does not map into domain 'reading'
Mar 31 19:35:22 ArchKnight rpc.idmapd[3979]: Client 0: (user) name "1503" -> id "99"
Mar 31 19:35:22 ArchKnight rpc.idmapd[3979]: Client 0: (group) name "512" -> id "99"
Mar 31 19:35:24 ArchKnight rpc.idmapd[3979]: New client: 2
Mar 31 19:52:04 ArchKnight rpc.idmapd[3979]: Stale client: 2
Mar 31 19:52:04 ArchKnight rpc.idmapd[3979]:    -> closed /var/lib/nfs/rpc_pipefs/nfs/clnt2/idmap

Oh yeah, LDAP is involved too (but it wasn't the problem so we'll ignore that detail). 

Solution:

So I was trying different things in the wicd postconnect/predisconnect scripts and it didn't seem like that was doing anything.
Then I went and checked the settings in the hibernate configuration (duh!) and found that it was still calling netfs to start/stop the network
and was mounting/unmounting nfs (not nfs4) filesystem types.

So once I fixed hibernate to stop wicd before suspend (and therefore shutdown network dependent servicees in predisconnect), and
adjusted the hibernate fs mount settings, all is well.