[mythtv-users] Stale NFS file handle

[Will Dormann]

Alan Anderson wrote:
> Hi,
> 
> A stale file handle occurs when the file handle cached by the nfs client is no 
> longer valid on the nfs server.   Something happened to the file it was 
> deleted moved to another file system or the NFS server was restarted.  The 
> only way out is to umount the file system on the nfs client and then mount it 
> again.  This should flush any cached file handles on the client.  Then the 
> client will get a new file handle from the nfs server that should be valid. 

Thanks for the info.

I think I have found a correlation with the mythbackend hanging and the 
fileserver.   On the fileserver, I have occasionally seen log lines like 
this:

Feb 23 21:22:54 [kernel] eth0: link down
Feb 23 21:23:02 [kernel] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
Feb 23 21:24:51 [kernel] eth0: link down
Feb 23 21:24:53 [kernel] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1
Feb 23 21:32:18 [kernel] eth0: link down
Feb 23 21:32:22 [kernel] eth0: link up, 100Mbps, full-duplex, lpa 0x45E1


It appears that under load, the Realtek 8139 (junk from what I hear) 
card doesn't fare so well?     I've just replaced the NIC with an 
eepro100, so we'll see how that works out.

What's curious is that I have seen the "link down" error messages on the 
NAS in the past, but it only seems that with MythTV 0.17 it makes a 
difference.   Also, I wonder which is better to use or possibly more 
robust...   NFS or SMB.


Thanks again.
-- 
-WD