Hi Vladimir,
On Fri, 2011-04-15 at 09:15 +0200, Vladimir Elisseev wrote:
> I'm pretty sure this is not the case for my environment, as we have
> less then 100 groups in total.
Re-reading the mail, I realize you might be right. The EINVALID on the
rpc channel file jumped out at me though since it was the same symptom
in our case. Also, in our case we saw the problem in all versions of
nfs-utils and not just the latest one, so it could be that it's an error
similar in nature but not the same underlying cause.
Anyway, in case it's any help (assuming that the patch doesn't fix the
problem for you), the kernel returns EINVALID on write()'s to the file
when any single write is not properly formatted and newline terminated.
It seems to be very picky about that. If you strace -s2048 the
offending daemon, it should be pretty clear though, either you'll see a
series of split up BUFSZ sized writes (each getting EINVALID) or you'll
see a write operation that's otherwise somehow improperly formatted.
Hope that's at least a bit helpful,
Sean
ÿô.nlj·Ÿ®‰†+%ŠË±é¥Šwÿº{.nlj·¥Š{±þwìíè jg¬±¨ ¶‰šŽŠÝjÿ¾«þG«é¸ ¢·¦j:+v‰¨Šwèm¶Ÿÿþø®w¥þŠà£¢·hšâÿ†Ù
