CIFSCLIENT daemon coredump

[Domenico Viggiani]

Hi,
I'm facing a serious problem with CIFSCLIENT.
A Win2K share (beh, really a high-availability share on a NAS Cluster...) is mounted by CIFSCLIENT on a HP-UX 11.00 machine.
Every long-running task on this filesystems aborts with a 'NFS/RPC timeout', leaving a stale mountpoint and a 'core' file in /var/opt/cifsclient/core directory.
I tried also with latest CifsClient version A.01.09.

Any idea?

[Darren Prior]

Hi,

Have you tried running a comparison on a Windows box? ie mapping the share and performing a similar task.

If the problem is only seen on CIFSclient, then I'd suggest you might need to tune some of the options. Full documentation for these is available under /opt/cifsclient/HP_Docs. The sections I'd concentrate on are: nfsTimeout and nfsRetransmit.

regards,

Darren.

[Steven E. Protter]

This might be a networking issue on the HP-UX box.

Please check out the following things.

lanadmin -x #

replace # with the 1 in lan1, whatever NIC is doing the connection.

If its not what you expect, it can cause the problem.

lanadmin -X 1 100FD

100 BaseT Full Duplex.

Check again.

If indeed these changes make the problem go away you can hard code your duplex settings in the file /etc/rc.config.d/hpbtlanconf

I am attaching an example file.

If you are going through a Cisco switch, have the HP-UX port setting set explicitly to 100 BaeT Full Duplex Manual if your NIC is not a 1000 BaseT NIC.


The above solution has solved CIFS/NFS and other networking issues for me in the past.

SEP