Hi Dave,

I guess you need to tcpdump to see what is really happening, but (one
of) the standard things to try for nfs hangs is to decrease the
read/write sizes. try adding something like ",-r=4096,-w=4096" to the
options list of client side fstab.

Best regards,

        -- Urban

>>>>> Dave  writes:

Dave> I have /usr/pkgsrc exported via NFS like this from /etc/exports:
Dave> /usr/pkgsrc -maproot=root -network 10.0.0.0/24

Dave> The client (10.0.0.3) mounts it like this from /etc/fstab:
Dave> arthur:/usr/pkgsrc /usr/pkgsrc nfs rw,noauto,nosuid,-i 0 0

Dave> /etc/rc.conf on both machines has mountd, nfs_server,
Dave> nfs_client, lockd, statd, and rpcbind set to YES.  The client is
Dave> set to do builds in /usr/tmp, so when a build is done, a symlink
Dave> is added instead of a directory.  Now, whenever a build decides
Dave> it needs to download something into /usr/pkgsrc/distfiles, the
Dave> write hangs with a process-in-disk flag.  From this point on,
Dave> any accesses to anything in /usr/pkgsrc from the client will
Dave> also hang with a process-in-disk flag.  Same goes for umount
Dave> attempts.  Interestingly, rpcinfo -p client when done from the
Dave> client also hangs.

Dave> -- David Griffith dgriffi@cs.csubak.edu