Re: Creating a Second IP Stack

2011-03-09 Thread Jeff Gribbin
Brilliant feedback - thanks - it's good to see that this has blossomed in
to
something that is at-least potentially useful to more folk than just myse
lf.

Just to update my personal progress, I've successfully built the second
stack and then, because I was able to now access the system through the
second stack, I've been able to configure a VSWITCH using what was the
primary stack's OSA interface and get the primary stack running through t
he
VSWITCH.  Happiness.

I didn't know about the OSA's internal stack - sounds like a useful
capability but not something that I'm personally in a position to current
ly
take advantage of.  The same goes for SYSG access - not feasible on my
current administrative setup but I'm glad that it's been mentioned.

Dumb question: If one logs on to the OSA's internal stack, what does the 
SCP
actually see?  Is it a local 3270 at some predefined address? I'd certain
ly
be interested in learning a little more about this capability - could you

maybe point me at the, 'proper' documentation?

I have a small puzzlement in that I can't ping either of my stacks from t
he
other ... I guess that I'm going to be learning a little more about the
intricacies of TCP/IP and subnetting in order to get to the bottom of why

THAT is the case - but not today, the sun is shining and the dogs are
telling me that it's a good day for a walk :-)

Again, thanks for all the input.

Jeff


Re: Creating a Second IP Stack

2011-03-09 Thread Kris Buelens
An OSAICC present itself to the outside world as a 3270 TELNET server.  To
the host it looks as LOCAL 3270's, with indeed 4-digt addresses like
D000-D0CF

The customisation defines with what address a TELNET client is present to
the host.  In PCOMM, the TELNET end-user one defines as part of its
connection defintions not only the IP address of the OSA ICC, but also a
LUNAME
There are two choises

   - a LU Name can correspond to a single 4 digit address.  If a client is
   already connected, others must select an other LUname
   - a LU Name can correspond to a list of addresses (then termed POOL
   Name),

Often one predefines some single address LU names, to connect operator
consoles etc. So soem workstations present themselves to the host at
well-known 4-digits addresses.  At least one pool is then defined for
everyone else.
Another practice I often see is that one changes the PORT number of the OSA
ICC from 23 for example 1000 or 3270, or ... So it is a bit hidden from the
public, only the intimates can then use the (limited number of) sessions
an OSA ICC provides.  This defintion all happens via the HCM, you select the
channels of the CPC and then channel Advanced facilities.

2011/3/9 Jeff Gribbin jeff.grib...@gmail.com

 Brilliant feedback - thanks - it's good to see that this has blossomed into
 something that is at-least potentially useful to more folk than just
 myself.

 Just to update my personal progress, I've successfully built the second
 stack and then, because I was able to now access the system through the
 second stack, I've been able to configure a VSWITCH using what was the
 primary stack's OSA interface and get the primary stack running through the
 VSWITCH.  Happiness.

 I didn't know about the OSA's internal stack - sounds like a useful
 capability but not something that I'm personally in a position to currently
 take advantage of.  The same goes for SYSG access - not feasible on my
 current administrative setup but I'm glad that it's been mentioned.

 Dumb question: If one logs on to the OSA's internal stack, what does the
 SCP
 actually see?  Is it a local 3270 at some predefined address? I'd certainly
 be interested in learning a little more about this capability - could you
 maybe point me at the, 'proper' documentation?

 I have a small puzzlement in that I can't ping either of my stacks from the
 other ... I guess that I'm going to be learning a little more about the
 intricacies of TCP/IP and subnetting in order to get to the bottom of why
 THAT is the case - but not today, the sun is shining and the dogs are
 telling me that it's a good day for a walk :-)

 Again, thanks for all the input.

 Jeff




-- 
Kris Buelens,
IBM Belgium, VM customer support


Re: Creating a Second IP Stack

2011-03-08 Thread Alan Altmark
On Tuesday, 03/08/2011 at 02:51 EST, Kris Buelens kris.buel...@gmail.com 
wrote:
 An OSA ICC is great, but costs real money: the price of that OSA card, 
it will 
 be dedicated to its OSA ICC function.

To clarify, except for 10 Gb ethernet, all OSA cards have two chpids.  You 
assign ICC on a per-chpid basis.

Alan Altmark

z/VM and Linux on System z Consultant
IBM System Lab Services and Training 
ibm.com/systems/services/labservices 
office: 607.429.3323
mobile; 607.321.7556
alan_altm...@us.ibm.com
IBM Endicott


Re: Creating a Second IP Stack

2011-03-08 Thread Kris Buelens
Thanks Alan, I should have mentioned that too.

2011/3/8 Alan Altmark alan_altm...@us.ibm.com

 On Tuesday, 03/08/2011 at 02:51 EST, Kris Buelens kris.buel...@gmail.com
 wrote:
  An OSA ICC is great, but costs real money: the price of that OSA card,
 it will
  be dedicated to its OSA ICC function.

 To clarify, except for 10 Gb ethernet, all OSA cards have two chpids.  You
 assign ICC on a per-chpid basis.

 Alan Altmark

 z/VM and Linux on System z Consultant
 IBM System Lab Services and Training
 ibm.com/systems/services/labservices
 office: 607.429.3323
 mobile; 607.321.7556
 alan_altm...@us.ibm.com
 IBM Endicott




-- 
Kris Buelens,
IBM Belgium, VM customer support


Re: Creating a Second IP Stack

2011-03-08 Thread Ron Schmiedge
Also risking the wrath of Chuckie, I can tell you what the person on
IBMLink told me when I opened an ETR for this same request:

I added my second stack to SYSTEM DTCPARMS:

:nick.TCPIP2   :type.server
   :class.stack
   :attach.0C00-0C01

I created a TCPIP2 TCPIP file, which only does TN3270. It only has
port 23 open and it has its own DEVICE and LINK for the non-QDIO port
pair C00-C01 from our OSA-E, its own HOME address, but same mask and
gateway handed down from the mount by the Network Gods.

I created a TCPIP2 DATA file on 592 with the TCPIPUSERID set to TCPIP2.

And its been running that way for a couple of years.

I do use the TCP TCPIP2 option on NETSTAT commands.

IIRC, the TCPIP person who led me through this was musing that they
really should document how to do this in the manuals, since so many
people ask the same question. I have not gone to look at the recent
books.

Ron

On Sat, Mar 5, 2011 at 10:07 AM, Jeff Gribbin jeff.grib...@gmail.com wrote:
 Greetings,
 I may be a, 'VM-Oldie' but I'm most certainly a, 'TCP/IP Newbie' and, true
 to form, I'm already wishing to push the envelope a little ...

 I'm helping to maintain a not-very-busy z/VM system which is pretty-much
 starting from scratch and currently has no asociated SLA's or the such to
 worry about.  It's not quite a sandpit, but pretty close.  I access the
 system via TN3270 and, in order to allow me to perform surgery on the main
 TCP/IP stack (bog-standard vanilla configuration, straight out of the box),
 I'm looking to place a second IP stack alongside the primary stack that will
 give me at-least a TN3270, 'back door' into the system.

 My, 'problem' is TCPIP DATA, which appears to be a unique filename.  'TCP/IP
 Planning and Customization' states quite categorically that TCPIP DATA
 defines parameters used by TCPIP -client- applications.  If this is truly
 the case then it would seem that all I need in order to implement a second
 stack purely for TN3270 purposes is to create appropriate 'userid
 DTCPARMS' and 'userid TCPIP' files on TCPMAINT's 0198 and I'm away.

 Would I be correct in thinking that TCPIP DATA is, 'merely' read by the,
 'client' commands in order (e.g.) to locate the servers that they need to
 interact with. If this is so, would it be acceptable practise (that is,
 behaviour that would not bring the Wrath of Chuckie upon me) to keep a
 second copy of TCPIP DATA, configured according to the userid(s) of the
 second stack, on a separate minidisk and access that minidisk in front of
 TCPMAINT's 592 when wishing to, 'converse' with the second stack? (For
 example, in order to issue a NETSTAT command without needing to specify the,
 'TCP' option every time.)

 I can, of course, discover most of this empirically for myself - and will
 indeed be experimenting a little (what else to do on a wet Saturday
 afternoon?) - but if for no other reason that the Fear of Chuckie I would be
 reassured if somebody with more experience were willing to sanity-check my
 hypothesis and proposed, 'solution'.

 Thanks
 Jeff



Re: Creating a Second IP Stack

2011-03-08 Thread David Boyes
On 3/8/11 1:27 PM, Ron Schmiedge ron.schmie...@gmail.com wrote:


IIRC, the TCPIP person who led me through this was musing that they
really should document how to do this in the manuals, since so many
people ask the same question. I have not gone to look at the recent
books.

Would be a nice redbook-like thing -- something like A Practical Cookbook
of Networking Tasks for VM TCPIP. Adding things like getting a caching DNS
server running, setting up FTP servers with RACF and other ESMs,
configuring SMTP to forward all outgoing mail to a corporate gateway, etc,
etc. 

I'll see if we can write something up.

-- db


Re: Creating a Second IP Stack

2011-03-08 Thread Alan Altmark
On Tuesday, 03/08/2011 at 01:27 EST, Ron Schmiedge 
ron.schmie...@gmail.com wrote:
 Also risking the wrath of Chuckie, I can tell you what the person on
 IBMLink told me when I opened an ETR for this same request:
 
 I added my second stack to SYSTEM DTCPARMS:
 
 :nick.TCPIP2   :type.server
 :class.stack
 :attach.0C00-0C01

Good.

 I created a TCPIP2 TCPIP file, which only does TN3270. It only has
 port 23 open and it has its own DEVICE and LINK for the non-QDIO port
 pair C00-C01 from our OSA-E, its own HOME address, but same mask and
 gateway handed down from the mount by the Network Gods.

Good.

 I created a TCPIP2 DATA file on 592 with the TCPIPUSERID set to TCPIP2.

OK, but it's just taking up space since there is nothing to tell the 
socket functions to read TCPIP2 DATA.  Perhaps your TCPRUNXT is copying 
TCPIP DATA to the A-disk, overriding values with information from TCPIP2 
DATA?

 And its been running that way for a couple of years.
 
 I do use the TCP TCPIP2 option on NETSTAT commands.

The TCP operand overrides what is found in TCPIP DATA.


Alan Altmark

z/VM and Linux on System z Consultant
IBM System Lab Services and Training 
ibm.com/systems/services/labservices 
office: 607.429.3323
mobile; 607.321.7556
alan_altm...@us.ibm.com
IBM Endicott


Re: Creating a Second IP Stack

2011-03-07 Thread Tom Duerbusch
Where there are many reasons why you might want a second IP stack, if the only 
reason is to have a back door TN3270 session, then try using the ICC port.  (An 
OSA port configured for ICC.)  These are always up and you can TN3270 into 
them, get a VM logo (in the VM world), and fix what ever problems that messed 
up the stack.

An OSA port configured for ICC, has its own IP stack running in firmware.  You 
can't touch it.  You can only configure it.  Once configured, it's hard to mess 
up.  Certainly, you will never mess it up by changing the configuration on you 
IP stack (unless you try to assign the same IP address G.  But even if you do 
configure the same IP address, the first one up (the ICC) wins.)

Tom Duerbusch
THD Consulting

 Jeff Gribbinjeff.grib...@gmail.com 3/5/2011 10:07 AM 
Greetings,
I may be a, 'VM-Oldie' but I'm most certainly a, 'TCP/IP Newbie' and, true
to form, I'm already wishing to push the envelope a little ...

I'm helping to maintain a not-very-busy z/VM system which is pretty-much
starting from scratch and currently has no asociated SLA's or the such to
worry about.  It's not quite a sandpit, but pretty close.  I access the
system via TN3270 and, in order to allow me to perform surgery on the main
TCP/IP stack (bog-standard vanilla configuration, straight out of the box),
I'm looking to place a second IP stack alongside the primary stack that will
give me at-least a TN3270, 'back door' into the system.

My, 'problem' is TCPIP DATA, which appears to be a unique filename.  'TCP/IP
Planning and Customization' states quite categorically that TCPIP DATA
defines parameters used by TCPIP -client- applications.  If this is truly
the case then it would seem that all I need in order to implement a second
stack purely for TN3270 purposes is to create appropriate 'userid
DTCPARMS' and 'userid TCPIP' files on TCPMAINT's 0198 and I'm away.

Would I be correct in thinking that TCPIP DATA is, 'merely' read by the,
'client' commands in order (e.g.) to locate the servers that they need to
interact with. If this is so, would it be acceptable practise (that is,
behaviour that would not bring the Wrath of Chuckie upon me) to keep a
second copy of TCPIP DATA, configured according to the userid(s) of the
second stack, on a separate minidisk and access that minidisk in front of
TCPMAINT's 592 when wishing to, 'converse' with the second stack? (For
example, in order to issue a NETSTAT command without needing to specify the,
'TCP' option every time.)

I can, of course, discover most of this empirically for myself - and will
indeed be experimenting a little (what else to do on a wet Saturday
afternoon?) - but if for no other reason that the Fear of Chuckie I would be
reassured if somebody with more experience were willing to sanity-check my
hypothesis and proposed, 'solution'.

Thanks
Jeff


Re: Creating a Second IP Stack

2011-03-07 Thread Kris Buelens
An OSA ICC is great, but costs real money: the price of that OSA card, it
will be dedicated to its OSA ICC function.

An other backdoor, that costs you nothing, is using the Integrated 3270
Console, found on the HMC.  And if you setup remote access to the HMC, you
can use it from anywhere (like I do for example from Belgium to a z/VM
system in Switserland in emergency cases).  An Integrated 3270 Console has
its limitations though: no filetransfer, no graphics, not the classsic
PCOMM keyboard.  But for basic VM work it is perfectly OK:  FILELIST, XEDIT,
SAPL, all work perfectly well.

2011/3/7 Tom Duerbusch duerbus...@stlouiscity.com

 Where there are many reasons why you might want a second IP stack, if the
 only reason is to have a back door TN3270 session, then try using the ICC
 port.  (An OSA port configured for ICC.)  These are always up and you can
 TN3270 into them, get a VM logo (in the VM world), and fix what ever
 problems that messed up the stack.

 An OSA port configured for ICC, has its own IP stack running in firmware.
  You can't touch it.  You can only configure it.  Once configured, it's hard
 to mess up.  Certainly, you will never mess it up by changing the
 configuration on you IP stack (unless you try to assign the same IP address
 G.  But even if you do configure the same IP address, the first one up
 (the ICC) wins.)

 Tom Duerbusch
 THD Consulting

  Jeff Gribbinjeff.grib...@gmail.com 3/5/2011 10:07 AM 
 Greetings,
 I may be a, 'VM-Oldie' but I'm most certainly a, 'TCP/IP Newbie' and, true
 to form, I'm already wishing to push the envelope a little ...

 I'm helping to maintain a not-very-busy z/VM system which is pretty-much
 starting from scratch and currently has no asociated SLA's or the such to
 worry about.  It's not quite a sandpit, but pretty close.  I access the
 system via TN3270 and, in order to allow me to perform surgery on the main
 TCP/IP stack (bog-standard vanilla configuration, straight out of the box),
 I'm looking to place a second IP stack alongside the primary stack that
 will
 give me at-least a TN3270, 'back door' into the system.

 My, 'problem' is TCPIP DATA, which appears to be a unique filename.
  'TCP/IP
 Planning and Customization' states quite categorically that TCPIP DATA
 defines parameters used by TCPIP -client- applications.  If this is truly
 the case then it would seem that all I need in order to implement a second
 stack purely for TN3270 purposes is to create appropriate 'userid
 DTCPARMS' and 'userid TCPIP' files on TCPMAINT's 0198 and I'm away.

 Would I be correct in thinking that TCPIP DATA is, 'merely' read by the,
 'client' commands in order (e.g.) to locate the servers that they need to
 interact with. If this is so, would it be acceptable practise (that is,
 behaviour that would not bring the Wrath of Chuckie upon me) to keep a
 second copy of TCPIP DATA, configured according to the userid(s) of the
 second stack, on a separate minidisk and access that minidisk in front of
 TCPMAINT's 592 when wishing to, 'converse' with the second stack? (For
 example, in order to issue a NETSTAT command without needing to specify
 the,
 'TCP' option every time.)

 I can, of course, discover most of this empirically for myself - and will
 indeed be experimenting a little (what else to do on a wet Saturday
 afternoon?) - but if for no other reason that the Fear of Chuckie I would
 be
 reassured if somebody with more experience were willing to sanity-check my
 hypothesis and proposed, 'solution'.

 Thanks
 Jeff




-- 
Kris Buelens,
IBM Belgium, VM customer support


Re: Creating a Second IP Stack

2011-03-05 Thread Alan Altmark
1.  You are wise in your fear.
2.  You are correct in all your suppositions.

When I set up an alternate stack, I use TCPRUNXT or the :exit. tag to copy
TCPIP DATA from 198 to the 191 with appropriate changes.

Regards,

Alan Altmark
IBM Lab Services

-
Sent from my BlackBerry Handheld.


Re: Creating a Second IP Stack

2011-03-05 Thread Jeff Gribbin
:-)

(Exits I shall leave as an exercise for another day)