| Author |
Message
|
| iruntoo |
 Posted: Mon Apr 18, 2022 5:53 am Post subject: shutdown best practice |
 |
|
Newbie
Joined: 13 Nov 2013
Posts: 4
|
MQ v8 had a best practice to shutdown: issue endmqm -c
If the timer pops and the queue manager has processes left, issue endmqm -i, then endmqm -p.
With the advent of Cloud and faster shutdowns or vm moves. is this still the standard, or does mq v9 handle in a more proficient manner? |
|
| Back to top |
|
 |
| bruce2359 |
| Posted: Mon Apr 18, 2022 9:27 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
Are you referring to your application processes? Or to MQ internal processes?
After you issue endmqm -c command, and waiting a bit (like one minute), what app processes remain?
-c is controlled quiesce, which allows all currently connected apps to run to their normal disconnect, however long that takes, but no new connects.
Apps should include _FAIL_IF_QUIESCING on all MQI calls, and when the app receives the quiescing ReasonCode, the app should end itself as quickly as possible.
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| gbaddeley |
| Posted: Mon Apr 18, 2022 3:59 pm Post subject: |
|
|
Jedi Knight
Joined: 25 Mar 2003
Posts: 2538
Location: Melbourne, Australia
|
It depends on the behavior of your apps when the MQ queue manager shuts down. -c is the normal shutdown and can take less than a second. You may need to use to -i. This has never affected integrity in my experience.
_________________
Glenn |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Mon Apr 18, 2022 4:44 pm Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| gbaddeley wrote: |
| It depends on the behavior of your apps when the MQ queue manager shuts down. -c is the normal shutdown and can take less than a second. You may need to use to -i. This has never affected integrity in my experience. |
With a bit more clarity: endmqm in all three instances (-c, -i and -p), the qmgr itself ends normally. However, with -i and -p its your applications integrity will likely be affected.
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Tue Apr 19, 2022 4:20 pm Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
| bruce2359 wrote: |
| However, with -i and -p its your applications integrity will likely be affected. |
We have been shutting down hundreds of queue managers for decades using -i as the first choice in the endmqm script. "Application integrity" is not an issue.
Using -c is wishful thinking.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Wed Apr 20, 2022 4:40 am Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
| PeterPotkay wrote: |
| bruce2359 wrote: |
| However, with -i and -p its your applications integrity will likely be affected. |
We have been shutting down hundreds of queue managers for decades using -i as the first choice in the endmqm script. "Application integrity" is not an issue.
Using -c is wishful thinking. |
Perhaps "operational integrity" better describes the issue with -i and -p, which effectively prevent applications from running to their normal end-of-job within SLA This is an issue with most of my clients.
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| gbaddeley |
| Posted: Thu Apr 21, 2022 5:14 pm Post subject: |
|
|
Jedi Knight
Joined: 25 Mar 2003
Posts: 2538
Location: Melbourne, Australia
|
| bruce2359 wrote: |
| Perhaps "operational integrity" better describes the issue with -i and -p, which effectively prevent applications from running to their normal end-of-job within SLA This is an issue with most of my clients. |
In production, we would only ever shut down a queue manager if there was an approved change, with app signoff that there will be no impact to SLA.
Apps should be tolerant of MQ outages, with the ability to recover / re-run / reconcile as required. For MQ messaging, they should be able to rely on MQ QOS / commit. 
_________________
Glenn |
|
| Back to top |
|
|
| bruce2359 |
| Posted: Thu Apr 21, 2022 5:21 pm Post subject: |
|
|
Poobah
Joined: 05 Jan 2008
Posts: 9469
Location: US: west coast, almost. Otherwise, enroute.
|
_________________
I like deadlines. I like to wave as they pass by.
ב''ה
Lex Orandi, Lex Credendi, Lex Vivendi. As we Worship, So we Believe, So we Live. |
|
| Back to top |
|
|
| Andyh |
| Posted: Tue Jul 26, 2022 9:12 am Post subject: |
|
|
Master
Joined: 29 Jul 2010
Posts: 239
|
| The recent "-t" and "-tp" options added to endmqm allow a target shutdown time to be passed to the endmqm command. The queue manager will progressively escalate the shutdown type in an attempt to meet the target time. |
|
| Back to top |
|
|
| fjb_saper |
| Posted: Tue Jul 26, 2022 11:05 pm Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
| Andyh wrote: |
| The recent "-t" and "-tp" options added to endmqm allow a target shutdown time to be passed to the endmqm command. The queue manager will progressively escalate the shutdown type in an attempt to meet the target time. |
Not yet documented in the command help (-?), but documented in the infocenter. The command does accept -t. Can we please also get the -t / -tp documented in the command help/usage ?
Thanks 
_________________
MQ & Broker admin |
|
| Back to top |
|
|
|
|
