Yes, we think you are right, it must have to do with the os-release: we have
no problems with nv6 and aix 4.2.x, but we do have a problem with aix 4.3.x.
So - it seems not to be a "disaster" to have a "big" hosts file (like ours
with appr. 6000 entries) since we never had problems in aix4.2.x.

We will keep on testing and investigating - but - since it's now past
midnight - we'll have a little break ... ;-)

cordially

Herbert Adamczyk

INFORMATIONS-TECHNOLOGIE AUSTRIA GMBH
Netzwerk-Management

A-1020 Wien, Lassallestraße 5
Telefon: ++43-1-21717-58943
Telefax: ++43-1-21717-58979
mailto: herbert.adamczyk AT it-austria DOT com


-----Ursprüngliche Nachricht-----
Von: Ray Schafer [mailto:schafer AT tkg DOT com]
Gesendet am: Donnerstag, 15. Juni 2000 23:34
An: IBM NetView Discussion
Betreff: Re: [NV-L] netmon and /etc/hosts ?

Alfred,

This is just too interesting.  You are probably right about fgets and
/etc/hosts.

How do you get this stack trace, and how do you know it's hung in those
routines?  When I run dbx, it stops the program and then I can look at the
trace, so all I get are snapshots.  Is this one such snapshot?

TKG has a nice tool called sctrace.  You could get a trial version that will
show you the last 30 system calls.  It shows more details than dbx does, and
it
can show you the timings and arguments as well.  30 lines ain't much but it
may
be all you need in this case.

http://www.tkg.com/sc

I think it is interesting that it works fine after you clear the snmpconf
cache.

Note that the stack trace gets to the ho_get* calles from
_OVsnmpConfLookupDestData.

I seem to recall that there were some issues with IPv6 and AIX 4.3 (maybe
4.3.2) that talk about using bind3 or some such.

I think you need some really good tracing tools for this problem.  Maybe not
clearing the cache will cause a bunch of host lookups to occur and that's
what
is screwing you up.



--
Ray Schafer                   | schafer AT tkg DOT com
The Kernel Group              | Distributed Systems Management
http://www.tkg.com