My systems use the unqualified host name as 'hostname' -- so the
'hostname' command returns 'box' and not 'box.foo.com'. I don't remember
WHY I set them up this way - IIRC, it was SAP, HACMP, or PSSP on the SP
frame that required hostname be the 'shortname' AND eight or fewer
characters (this was lat '96, early '97, AIX 4.1, SAP/R3 3.6, HACMP 2.1,
and God only knows which HACMP version).

We're now trying to install some awfull java-based functionality for SAP
and it fails -- can't connect to 'box'. If we change to a fully qualified
hostname, it seems to work. It looks like it does a lookup on 'box', gets
1.2.3.4, does a reverse on 1.2.3.4 and gets 'box.foo.com' and says this
isn't 'box'.

I can't seem to find a reference to 'shortname' anywhere, and I'm
wondering how big a can of worms we'll need to re-can if we start changing
the hostname across our production landscape.

What am I missing here? Am I between the proverbial rock and hard place?

We're on AIX 5.1 ML 4, HACMP/ES 4.5.1, SAP 4.6, and no longer an SP
environment.

Oh, and for what it's worth, SAP indicates that hostname needs to be fully
qualified . . .

Tom

Take a look at /etc/netsvc.conf - you should be able to specify the lookup
order (ie - hosts=local,bind)

"TomK" <namffuak@NO.skyenet.SPAM.net> wrote in message
news:0rhlv0l4m8na1gp0jsqphs9rs6fi7n3b8t@4ax.com...
> My systems use the unqualified host name as 'hostname' -- so the
> 'hostname' command returns 'box' and not 'box.foo.com'. I don't remember
> WHY I set them up this way - IIRC, it was SAP, HACMP, or PSSP on the SP
> frame that required hostname be the 'shortname' AND eight or fewer
> characters (this was lat '96, early '97, AIX 4.1, SAP/R3 3.6, HACMP 2.1,
> and God only knows which HACMP version).
>
> We're now trying to install some awfull java-based functionality for SAP
> and it fails -- can't connect to 'box'. If we change to a fully qualified
> hostname, it seems to work. It looks like it does a lookup on 'box', gets
> 1.2.3.4, does a reverse on 1.2.3.4 and gets 'box.foo.com' and says this
> isn't 'box'.
>
> I can't seem to find a reference to 'shortname' anywhere, and I'm
> wondering how big a can of worms we'll need to re-can if we start changing
> the hostname across our production landscape.
>
> What am I missing here? Am I between the proverbial rock and hard place?
>
> We're on AIX 5.1 ML 4, HACMP/ES 4.5.1, SAP 4.6, and no longer an SP
> environment.
>
> Oh, and for what it's worth, SAP indicates that hostname needs to be fully
> qualified . . .
>
> Tom

On Fri, 28 Jan 2005 19:49:11 -0500, "JSimth" <jsmith@microsoft.com> wrote:

>"TomK" <namffuak@NO.skyenet.SPAM.net> wrote in message
>news:0rhlv0l4m8na1gp0jsqphs9rs6fi7n3b8t@4ax.com.. .
>> My systems use the unqualified host name as 'hostname' -- so the
>> 'hostname' command returns 'box' and not 'box.foo.com'. I don't remember
>> WHY I set them up this way - IIRC, it was SAP, HACMP, or PSSP on the SP
>> frame that required hostname be the 'shortname' AND eight or fewer
>> characters (this was lat '96, early '97, AIX 4.1, SAP/R3 3.6, HACMP 2.1,
>> and God only knows which HACMP version).
>>
>> We're now trying to install some awfull java-based functionality for SAP
>> and it fails -- can't connect to 'box'. If we change to a fully qualified
>> hostname, it seems to work. It looks like it does a lookup on 'box', gets
>> 1.2.3.4, does a reverse on 1.2.3.4 and gets 'box.foo.com' and says this
>> isn't 'box'.
>>
>> I can't seem to find a reference to 'shortname' anywhere, and I'm
>> wondering how big a can of worms we'll need to re-can if we start changing
>> the hostname across our production landscape.
>>
>> What am I missing here? Am I between the proverbial rock and hard place?
>>
>> We're on AIX 5.1 ML 4, HACMP/ES 4.5.1, SAP 4.6, and no longer an SP
>> environment.
>>
>> Oh, and for what it's worth, SAP indicates that hostname needs to be fully
>> qualified . . .
>>
>> Tom
>
>Take a look at /etc/netsvc.conf - you should be able to specify the lookup
>order (ie - hosts=local,bind)
>

We are -- and the short name is listed first in /etc/hosts (HACMP again).
I'm about to go bomb-lobbing anarchist on these configs -- four
interfaces, with different IP networks, with IP (DNS) names of box1, box2,
box3, and box4 -- and a host name of box. There's too much crap out there
that associates AN IP address with THE host name when there are multiple
ways of reaching the system.

Tom

"JSimth" <jsmith@microsoft.com> wrote in
news:vP2dnTBrZL18HmbcRVn-rg@rogers.com:
> Take a look at /etc/netsvc.conf - you should be able to specify the
> lookup order (ie - hosts=local,bind)

Unless the DNS servers are correctly configured, the system will hang while
trying to reverse the IPv6 loopback address.

The fix to this is to use hosts = local, bind4 (Resolve only IPv4
addresses) this is no great loss if you're not using IPv6 yet.


--
Doing AIX support was the most monty-pythonesque
activity available at the time.
Eagerly awaiting my thin chocolat mint.

Tom,

Look at your /etc/resolv.conf and confirm
it has a "domain" stanza...
i.e.: domain mydomain.com

>From the man page on resolv.conf:
A domain entry tells the resolver routines which default domain
name to append to names that do not end with a . (period).
There can be only one domain entry. This entry is of the form:
domain DomainName

HTH,
-tony

On 07 Feb 2005 21:00:09 GMT, Jose Pina Coelho <eresquigal@netcabo.pt>
wrote:

>"JSimth" <jsmith@microsoft.com> wrote in
>news:vP2dnTBrZL18HmbcRVn-rg@rogers.com:
>> Take a look at /etc/netsvc.conf - you should be able to specify the
>> lookup order (ie - hosts=local,bind)
>
>Unless the DNS servers are correctly configured, the system will hang while
>trying to reverse the IPv6 loopback address.
>
>The fix to this is to use hosts = local, bind4 (Resolve only IPv4
>addresses) this is no great loss if you're not using IPv6 yet.

That's the option I was looking for this afternoon -- we just got the
report back from AIX support at SAP that the problem is that it's doing
the IPV4 lookup followed by an IPV6 lookup, with pointers to two APARS --
one to TCP, to take the first response, and one to the libc runtime to not
even do IPV6. I think we'll try this first.

And FWIW -- the problem is ONLY showing up on a system with HACMP 4.5.1
installed; it works as expected on two other identical software loads
minus HACMP.

Tom