I've set up an NFS server on a older system running Fedora 17 (kernel 3.9.10-100), and a client running Fedora 20 (kernel 3.19.8-100). My problem is that after manually mounting one share on the client, any other NFS operation will lock up the process calling that operation. For example, mounting the second share, executing ls on the share or the parent directory, etc. There is no reported error on the initial mount operation.
Server is 192.168.57.11
Client is 192.168.57.10
Both machines have a wireless connection to the internet, and a wired ethernet link to each other.
I have observed that when the process on the client has locked up, the server has often subsequently lost networking functionality. Not every time, and not immediately.
Server firewall is disabled.
Client firewall has wired interface in the 'trusted' zone with all services enabled.
server /etc/exports
Code:
/home/public 192.168.57.0/255.255.255.0(rw,insecure,sync,no_subtree_check,mp,all_squash,anonuid=1000,anongid=1000)
/home/dave 192.168.57.0/255.255.255.0(rw,insecure,sync,no_subtree_check,mp,all_squash,anonuid=1000,anongid=1000)
server /etc/hosts.allow
Code:
ALL:192.168.57.0/255.255.255.0
server /etc/hosts.deny
Checking with 'ps', the server has the appropriate instances of mountd, idmapd and nfsd running.
client /etc/fstab has the following lines
Code:
192.168.57.11:/home/public /home/public nfs rw,suid,exec,noauto,user,sync,hard,intr,timeo=14,rsize=8192,wsize=8192 0 0
192.168.57.11:/home/dave /home/dave/remote nfs rw,suid,exec,noauto,user,sync,hard,intr,timeo=14,rsize=8192,wsize=8192 0 0
Checking with 'rpcinfo -p 192.168.57.11', the client can see the ports for portmapper, nfs, mountd and lockd (among others).
Logged into client as user, in a bash window, execute 'mount /home/public'. Prompt comes back with no error. Then executing any of:
$ mount /home/dave/remote
$ ls /home/public
$ ls /home
will lock up that bash process.
Being logged in to the client as root makes no difference.
Once the bash process has locked up, another window still gets a ping response from the server. But performing network functions on the server, such as using a browser, can cause the server to become unavailable to pings from the client.
I've done all the troubleshooting I can find & understand, without attempting a stack trace. Any help or ideas will be appreciated.