What's curious is that moving the source trees over to a local
filesystem results in perfectly fine executables that have no problems
linking. Does anyone know why this occurs, or how to fix it?
There are also the "can't silly-delete" about every other second (yes
that's exaggeration; it occurs reproducibly after I EOF out of an `su'),
and lots of "__nfs_fhget: inode 536457 still busy, i_count=2" followed
by "__nfs_fhget: killing filehandle", which usually precipitates a ton
of "nfs_dentry_deletes" all over the place. I can provoke more of them
when this happens by doing a find on the tree which houses the
file...there is one for each in the tree.
Oh and one more...programs that use fcntl() cannot get a lock on the
files, and this results in things like MUAs being unable to open
mailboxes RW. The error says that `lockd' failed to monitor the host on
which the FS resides physically. The lockd does seem to be running
though, on both sides of the mount...
The mounts are from Solaris boxes running SunOS 5.5.1. Mount options
are rw,rsize=8192,wsize=8192,hard. Kernel is 2.2.10.
Does anyone know how to fix these problems? Would using HJL's knfsd
patches, even though I'm only a client, fix any of these? I am grateful
for assistance in this matter; we were going to use Linux here at work
for arch-exporting /usr/local to our various platforms but the Linux NFS
code seems highly immature at this point and we're probably better off
using one of the commercial UNIX offerings for this, unfortunately (and
patching the kernel every day is obviously out of the question).
Thanks.
-- Scott- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/