Our setup is similar to Marcy's.  Each SSI cluster has a master list of guest 
userids and IP addresses in a CMS file.  Each entry contains production and DR 
test IP addresses.  We used to have separate addresses for real DR, but those 
are now the same as production.  The IP addresses are already registered in DNS.

The program that builds guests takes the next available name from the master 
list, updates the list to show that the name is in use, and builds the VM 
userid.  It also builds a "guestname CONFIG" file on the shared 191 minidisk 
that all guests use.  When Linux boots, it reads this configuration file from 
its 191 and configures the network.  When a z/VM system is IPLed in DR test 
mode, a program updates the guest configuration files with IP addresses from 
the master list.

Like Marcy said, I don't see anything to gain by involving the DHCP people.  
Managing MACID's is no less complicated than managing IP addresses.

                                                                                
                                                            Dennis

"I'm not a football expert."  -- San Francisco 49ers CEO Jed York

-----Original Message-----
From: Linux on 390 Port [mailto:LINUX-390@vm.marist.edu] On Behalf Of Marcy 
Cortes
Sent: Friday, January 22, 2016 15:52
To: LINUX-390@vm.marist.edu
Subject: Re: SSI guest best practice

Alan wrote:

>I am wondering if people are looking at using persistent DHCP for this. To 
>make it work you have to use layer 2 VSWITCHes and you have to manage 
>MACIDs on the NICDEFs.   Your automation tools have to "manufacture" a new 
>MACID for every new NICDEF and put it there or on COMMAND DEFINE NIC. 
>(I've created such automation for DIRMAINT.)

>It means that the network people can operate the DHCP server, whether its 
>onboard as a Linux guest or outboard, and take responsibility for changing 
>things (home, DR, IP changes, etc.) 

I'm not.  I don't want to have to involve another group involved.   Right now 
DR is entirely automated and operations staff does the whole thing.   Our VM 
ids map to IP addresses (not hostnames) and I can relate the 2 without looking 
at a spreadsheet all the time.   On servers with more than one interface their 
IPs end in the same octet.  I don't want network people sticking other devices 
in our subnets and messing with the scheme :)  And i don't want to have to 
submit (more) paperwork!  VM provided MACIDs are fine with us.

Marcy

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

----------------------------------------------------------------------
This message, and any attachments, is for the intended recipient(s) only, may 
contain information that is privileged, confidential and/or proprietary and 
subject to important terms and conditions available at 
http://www.bankofamerica.com/emaildisclaimer.   If you are not the intended 
recipient, please delete this message.

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

Reply via email to