Question about NFS mounting mailspool

Question about NFS mounting mailspool

Post by Griff Mill » Sat, 07 Jun 1997 04:00:00



The Sun 4.x FAQ says that you should mount the mailspool with noac. Is
this the same as actimeo=0 ?

Thanks.

--
Griff Miller                              "I don't even put bumperstickers on
Senior Unix Sysadmin   '85 VF1100S         my car; why would I get a tattoo?"
PGS Tensor             '95 Z-28 6 spd.

 
 
 

Question about NFS mounting mailspool

Post by Robert Thurl » Wed, 11 Jun 1997 04:00:00




>The Sun 4.x FAQ says that you should mount the mailspool with noac. Is
>this the same as actimeo=0 ?

Not quite, in practice.  "noac" says never to cache attributes,
while "actimeo=0" says don't trust cached attributes.  They are
almost the same, but "actimeo=0" means you can likely get through
a complete logical operation without refetching attributes, while
"noac" will refetch them several times at different places in the
Solaris code for some operations.  We'd advise "actimeo=0".

Rob T
--

There was something fishy about the butler.  I think he was a Pisces,
probably working for scale.             -- Nick Danger, Third Eye

 
 
 

1. Goofy Results with NFS Mailspool

For many years, we have have used the NFS automounter to mount our
shared mailspool on all our systems.  The mail server is an RS/6000
running
AIX 4.3.2, with a variety of (mostly Sun Solaris) clients.

I have six essentially identical Solaris 7 x86 (Intel) machines, all on
the
same subnet, and all at essentially the same operating system and
patch level.  Three of them have no trouble at all accessing the mail
spool; the other three get messages like this when a user tries to
access
mail:

Couldn't create lock file /var/spool/mail/tkevans.lock! [Value too large
for defined data type]

Looking on the RS/6000 mail server, I see this in the spool directory:

# cd /var/spool/mail
# ls -l tke*
-rw-rw----   1 tkevans  mail      515936 Jul 13 16:08 tkevans
----------   1 tkevans  mail           0 Dec  6 1953  tkevans.lock
#

No kidding--1953!  While this seems to mean I have even longer than
most to worry about Y2K, this goofiness prevents users from accessing
their mail, and the funny lock file has to be manually removed on the
mail
server.

None of my Sparc Solaris machine (in the same subnet) have this problem,
and neither do my SGI IRIX systems or HP-UX systems.  Only these three
x86 Solaris machines do; and three other identical x86 Solaris machines
on
the same subnet don't!

Anyone have any ideas on this?
--
Tim Evans                     |    E.I. du Pont de Nemours & Co.

(302) 695-9353/8638 (FAX)     |    P.O. Box 80357

2. Local Printer

3. /var/mail: to NFS mount or not to NFS mount?

4. SCO Technical Articles released on 20000212

5. nfs mount to a nfs mount fails

6. Mouse Systems PS/2 Point Pro problems

7. nfs mount problem: mount: can't get address for nfs-server.kauai

8. How to upgrade libc(libg++)? --make doesn't work after upgrading

9. NFS mount won't mount at boot, but mount -a works fine.

10. amd-utils/nis/solaris/nfs-automounting question.

11. Frequency of NFS lock problems w/ NFS mounted mail spool

12. linux NFS server problem - can't ls in the NFS mounted directory.

13. NFS mount question (OpenBSD 2.8).