Bug 431623

Description of problem:

I have a new 64-bit RHEL5.1 server connected to two RAIDs via fiber, and a
network of 32-bit Fedora 7 workstations. We use a separate Fedora Directory
server to handle the authentication. This was all working until we installed the
new RAID and file server.
  Periodically the NFS will freeze. Checking the log I will see:

rpc.idmapd: nfsdcb: id '-2' too big! 

  When I see that, no new NFS connection will happen, and the existing ones will
freeze within a few minutes.

  Here's my /etc/exports:

/export
192.168.1.0/24(ro,fsid=0,insecure,no_subtree_check,sync,anonuid=65534,anongid=65534)
/export/d0
192.168.1.0/24(rw,fsid=1,nohide,insecure,no_subtree_check,anonuid=65534,anongid=65534)
/export/d1
192.168.1.0/24(rw,fsid=2,nohide,insecure,no_subtree_check,anonuid=65534,anongid=65534)

  We are using autofs to mount the drives. Here the file:

d0      -rw,fstype=nfs4,bg,soft,intr,noatime,proto=tcp   192.168.1.8:/d0
d1      -rw,fstype=nfs4,bg,soft,intr,noatime,proto=tcp   192.168.1.8:/d1

  And lastly, here's the idmapd.conf file:

[General]

Verbosity = 3 
Pipefs-Directory = /var/lib/nfs/rpc_pipefs
Domain = berkeley.edu

[Mapping]

Nobody-User = nfsnobody
Nobody-Group = nfsnobody

[Translation]
Method = nsswitch

Version-Release number of selected component (if applicable):


How reproducible:

  Sometimes it takes a few hours for this to happen. Other times it happens
almost immediately after restarting NFS.

Steps to Reproduce:
1.
2.
3.
  
Actual result:
NFS freezes.

Expected results:
It works.

Additional info:

  We read the ticket id=225507 and it didn't solve the problem.