Quote:
Originally Posted by jggimi
Does the server have multiple network routes to the client? I'm guessing, of course, because you have not drawn us any diagrams.
You might try mount_nfs with -T ... If that functions better for you look to a network topology root cause.
I'd also be interested in the rest of your custom sysctls, and ask if you've used the diagnostic tools described in FAQ 6.7.
|
The topology is the most simple at all - OpenBSD as a router and clients behind, all on the same network, connected to one and the same switch.
I figured out that from 4.9 on I don't need any network options tweaking in sysctl.conf, so there is only
net.inet.ip.forwarding=1 in there.
The problem with diagnostic tools is that they don't reveal much.. rpcinfo shows all services are runnig fine (they must as I can mount and ls), nfsstat shows some error probably
Server Ret-Failed 10
Server Faults 0
NFS over TCP seems to help only that much it doesn't hang the client and the directory can be unmounted, but I believe that depends more on the client