Re: Do NOT do ENDSBS without shutting down TCP first!

I believe this was indeed your issue.

For decades now we've look at line descriptions to assure they are NOT using a hard coded MAC address, ESPECIALLY during migrations. I can tell tales of migrations that were nearly a disaster due to these coded addresses.

Remember that back in the days of SNA it was important to know the mac address of your system, and it got plugged into many clients. As a consequence NOT changing it when you moved to a new server was a VERY attractive migration option.

But if you're using TCP/IP, and you ARE using TCP/IP, knowing the mac address is only important during troubleshooting.

DO BE AWARE that you CAN have TCP/IP running and an interface up and running even in restricted state! This has been true at least back to i 7.2, however ENDSBS *ALL or ENDSYS WILL take those addresses down.

- DrF

On 3/4/2023 7:18 PM, Rob Berendt wrote:

It may have been more of a MAC address issue. The line description on both
the old and the new had the same mac addresses. Remember to vary those
lines off...

On Sat, Mar 4, 2023 at 7:10 PM Rob Berendt <robertowenberendt@xxxxxxxxx>
wrote:

Do NOT do ENDSBS without shutting down TCP first!

I've ran into issues in the past. One time was so drastic that it took a
lot of work to get 5250 to work again.

However, this time it was on an old lpar that I wanted to leave up, but in
restricted state in case I needed to compare stuff.
I actually brought it out of restricted state before firing up the new one
so I could print off the brms recovery report.
I probably skipped the ENDTCP stuff and jumped straight to ENDSBS *ALL.
Figuring if it damages stuff, what do I care? I burn the boats when
migrating.

Result was dueling IP addresses from the old/new system. Even though
WRKSBS, F11 showed QCTL as "RSTD". And you cannot see the interface status
or do ENDTCPIFC when the system was brought into restricted state like this.
Varying the lines off did it.

This is not OS400. Do not treat it as such. Treat it like IBM i. IBM
cautions you greatly about running ENDSBS without ending tcp first.