| Author |
Message
|
| PeterPotkay |
 Posted: Thu Feb 26, 2015 7:49 am Post subject: When are Database connections released for addit' instances |
 |
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
http://www-01.ibm.com/support/knowledgecenter/SSKM8N_8.0.0/com.ibm.etools.mft.doc/ac00406_.htm?lang=en
| Quote: |
All other database connections remain open until one of the following events occurs:
An error occurs, while accessing the database, that requires a new connection to be made.
The message flow has been idle for 1 minute.
The message flow is stopped.
The broker is stopped. |
If a flow has Additional Instances configured to lets say 100, what constitutes an idle message flow I wonder.
If the message flow gets busy, and 101 connections are established to the database, and then the flow gets less busy so only 10 or so instances are needed, should we expect the 91 extra DB connections to be closed after a minute?
Or as long as even one instance remains active, then the flow is not considered idle, and so ALL connections remain, meaning we have 101 connections even though only 1 is needed?
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
 |
| mqjeff |
| Posted: Thu Feb 26, 2015 7:53 am Post subject: Re: When are Database connections released for addit' instan |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
| PeterPotkay wrote: |
| should we expect the 91 extra DB connections to be closed after a minute? |
Yes, as far as I know. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Thu Feb 26, 2015 8:30 am Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
That was our assumption too. But dealing with complaints from the Database dudes about stale connections from WMB.
Reading that sentence again, you could interpret it to mean that if even one instance of the flow is still active, then the flow is not considered idle, so all connections that the flow has made are maintained.
We will test by put inhibiting the input queue while all new work goes to the other cluster input queues. After 1 minute, that flow on that Broker should truly be idle, and we assume / hope ALL 101 connections get cleaned up. That will at least prove the theory.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
| mqjeff |
| Posted: Thu Feb 26, 2015 8:35 am Post subject: |
|
|
Grand Master
Joined: 25 Jun 2008
Posts: 17447
|
Well, here's the real question.
Does it mean that the database connection has been idle for one minute?
Or does it mean that the instance of the message flow has been idle for one minute AND THEN the database connection has been idle for ANOTHER minute?
Or does it mean that all instances of the flow have been idle for one minute... which is not how I understood it. |
|
| Back to top |
|
|
| mgk |
| Posted: Thu Feb 26, 2015 8:35 am Post subject: |
|
|
Padawan
Joined: 31 Jul 2003
Posts: 1642
|
It will depend on your version - there were some issues with DB connections from additional instances not expiring after the expiry time, which have since been fixed I believe. I can't remember the version that contains the fix though so you should raise a PMR to find out if your test proves they are not being released...
Kind regards,
_________________
MGK
The postings I make on this site are my own and don't necessarily represent IBM's positions, strategies or opinions. |
|
| Back to top |
|
|
| PeterPotkay |
| Posted: Thu Feb 26, 2015 10:11 am Post subject: |
|
|
Poobah
Joined: 15 May 2001
Posts: 7722
|
We are running 8.0.0.3 with plans to go to 8.0.0.5. I looked at the fixes in 8.0.0.4 and 8.0.0.5 and nothing jumped out at me.
I will update this post with our test results.
_________________
Peter Potkay
Keep Calm and MQ On |
|
| Back to top |
|
|
|
|
