Network issue, losing inetd services, ssh, & nfs mounts.

Network issue, losing inetd services, ssh, & nfs mounts.

Post by Quasimo » Fri, 03 Oct 2003 06:10:43

I need a little troubleshooting advice.
In a nutshell...
My server, a e450 running Solaris 5.6 is losing network connectivity.
SunOS foobarserver 5.6 Generic_105181-26 sun4u sparc SUNW,Ultra-4

Setup: e450 Solaris 5.6, hme0 & qfe0 have accessible IP's.
Symtoms: I'm ssh'd into both IP's tailing log files, I'm telneted in
to both IP's, tailing log files & ping tests, and I get dropped
randomly 45 seconds to sometimes 2 minutes, zero to 2 times a day.
rlogin & ftp die as well. The server becomes unpingable from the
outside using either IP.
NFS mount is mounted as follows;
/usr/stc from foobarserver2:/export/stc
Flags: vers=2,proto=udp,sec=sys,hard,intr,dynamic,rsize=8192,wsize=8192,retrans=5
 Lookups:     srtt=7 (17ms), dev=3 (15ms), cur=2 (40ms)
 Reads:     srtt=4 (10ms), dev=3 (15ms), cur=2 (40ms)
 Writes:     srtt=7 (17ms), dev=4 (20ms), cur=2 (40ms)
 All:     srtt=7 (17ms), dev=3 (15ms), cur=2 (40ms)

I understand that version2/udp isn't very good, but the netware server
won't accept a vers3/tcp connection as it needs to be upgraded.
/var/adm/messages is pummelled with nfs timeouts for 1 second & back
the same second. The outages I'm experiencing can be correlated to
those nfs timeouts which span 30-120 seconds. I was initially thinking
that having this NFS mount under /usr probably wasn't the best idea as
it locks up the /usr directory (not sub directories) & thought it may
be causing our problems, i.e. the cause, not the effect, but I have
been unable to reproduce it in the lab & ssh is running out of /opt.
Would not being able to stat the /usr directory in any way be causing
this problem? I'm installing the latest recommended patches this
Sunday, what else should I be looking at?
The NFS errors are the only thing showing up in /var/adm/messages.
Today I had a similar problem with a development server, the nfs
mounts are tcp/vers3, same kind of outage, no messages (probably
because its version3 & not udp/2). So now I'm back to thinking it's a
network issue.


1. any /etc/inetd.conf services connected with NFS


I've cut off a number of services I didn't think I needed from

NFS now is not working on the *export* side... it is working on the *import*

Any clues appreciated..... I've got 2 machines setup I thought identically
and only this one isn't exporting NFS-wise.



My FROM address is fake.  It does not exist.  It never has.  It is
not even currently possibly real.  I'll check back for followups.
This is the ``least-worst'' solution to dealing with spammers.  

2. CDR/RW still looking

3. NFS mounting & network locking

4. User security

5. Nfs-mount & RPC/network hanging/nntp inews

6. Digiboard PCx/e

7. Calling programs through inetd & services

8. ppp connection

9. Are sendmail & inetd services necessary?

10. ADSL, ssh & X (Was: ssh & X)

11. NFS cant mount from one network, mounts from another fine???

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

13. SUMMARY: NFS mounting /var/spool/mail && file locking