MVSFORUMS.com

[loblesa]

Good afternoon,

I have a project to perform the stop process of the systems and I desire that you recommend me about the order of it...

We use a rexx program to make this stopping sequence...

[taltyman]

No 2 shops are totally alike, so no one can give you specifics for your shop.
Generally you reverse the order that things are started when you shut down. At some point what is left up can happily survive an IPL. Actually everything should survive a hard down since eventually it will happen whether it is intentional or not.
Most shops stop the initiators, start bringing down the online systems such as CICS as well as TSO. Near the end you shut down your communication servers such as NET and TCPIP. You can bring down just about everything if you want to or you can just shutdown the finicky stuff and let the IPL handle the rest. It's a question of how much time you want to use to shutdown your system.

[warp5]

We shut down almost everything except TSO and VTAM/TCPIP. This way we can make necessary changes or watch the system before we have to leave our Terminal. Then we can shut down those last few components from the console.

[loblesa]

Thanks Taltyman and Warp5 for your response

Of course that every installation is diferent but for example...

Is necessary stop LLA and VLF? We make always IPL with CLPA...

Is correct to cancel the SMF?


Chr

[Mervyn]

I always wondered, do you watch for the little white dot in the centre of the screen to fade away to nothing?



_________________
The day you stop learning the dinosaur becomes extinct

[taltyman]

I'd leave VLF and LLA up. SMF may depend on how you switch/process your SMF datasets. But then I'm not a patient person when it comes to an IPL.

[warp5]

We do not cancel SMF, I do not think there is a reason to do that. We have a proclib member STOPSYS1 that stops almost everything. Then we have a STOPSYS2 that stops the last components. Then there is just a $P JES2,TERM and a Z EOD to finish the shutdown.
This is what our STOPSYS2 looks like:

[loblesa]

TaltyMan, about SMF... if we cancel SMF... could we lost records? I don't find the difference between cancel smf and stop smf... do you know it?

Thanks

[loblesa]

warp5, we have two scenarios...

1) systems without GDPS: we have automated the sequence of stop system even '$PJES2' command... stop JES2 and Halt EOD are done handle... but 'Z EOD' not always is done... is important to instruct operations team to do it? one of the most important steps of 'Z EOD' is to copy the cached data in 3990 devices to DASD... although it store the internal I/O device error counts in the logrec data set, empty the smf buffers and close the system log and put in on the print queue...

2) systems with GDPS: GDPS controls all the stop system process due to it make '$PJES2' and 'Z EOD'...


One year ago, I'm a operator in another business and we had done always 'Z EOD' command... it was a rule...


Thanks

[taltyman]

I don't think we ever bring down smf.
sto(P) is generally the first choice to bring down a started task, (C)ancel is a bigger hammer while the command of last resort is (F)orce

[taltyman]

If you read up in the MVS commands manual about the halt (Z) command you will find your answer on SMF.

[loblesa]

taltyman, about halt comand in the MVS commands manual I have seen that...

ii) Empty the SMF buffers onto the active SMF data set in SYS1.MANx
iv) Switch to another SMF data set in SYS1.MANx, allowing the previously active SMF data set to be dumped


then... is necessary stop the SMF?? I think that with 'Z EOD' command it shouldn't necessary stop the SMF... is correct?

Thanks a lot

[warp5]

You do not have to stop SMF. We do not. If you stop SMF how are you going to do the Z EOD?

[loblesa]

Yes warp5, you are right

Very thanks