:> We all owe Matt big for this. :)
:>
:>
:> > I'm using the default mount operations, as far as NFS server
:> > not responding messages, i have no clue, but the server is still
:> > up and i've seen that message happen when a lot of pressure is
:> > being put on an NFS server even though everything is fine.
:>
:> Try mounting with -d... Can I make a guess that the NFS mount is going over
:> 100MB ethernet? I have a strong theory that the dynamic retransmit timer
:> needs rework for low latency connections, with high variability in their
:> performance during high traffic. (lots of collisions)
:
:Hmmm, well the client/server are both on 100mbit
:full duplex switched ethernet, fxp0<->fxp0 with no load
:other than the build going on.
:
:-Alfred
I think Kevin is right re: the dynamic retransmit timeout. I've
noticed it occuring on my systems too -- although without the lockups.
But Kevin has other problems w/ his network that could be contributing
to the problems he is experiencing.
:btw, just finished run 4 of buildworld, still kicking.
I've gone through 4 buildworlds so far without any errors using an
NFS mounted /usr/src and /usr/obj. Though, I must say, it takes
about twice as long as with a FFS+softupdates /usr/obj ( two hours verses
50 minutes ). The read optimization is very good, though.
Matthew Dillon
<[email protected]>
To Unsubscribe: send mail to [email protected]
with "unsubscribe freebsd-current" in the body of the message
