| Author |
Message
|
| friedl.otto |
 Posted: Wed Feb 06, 2008 5:37 am Post subject: Lengthy ... but interesting channel config question. |
 |
|
Centurion
Joined: 06 Jul 2007
Posts: 116
|
Textbook SDR/RQSTR Example 
On QMGR_A running on hostA.domain.com:4500
| Code: |
DEFINE QLOCAL('QXMT_A_B') DEFPSIST(YES) MAXDEPTH(10000) USAGE(XMIT) REPLACE
DEFINE CHANNEL('CHNL_A_B') CHLTYPE(SDR) CONNAME('hostB.domain.com(9000)') XMITQ('QXMT_A_B') REPLACE
DEFINE QREMOTE('QRMT_DATA') RQMNAME('QMGR_B') RNAME('QLCL_DATA') XMITQ('QXMT_A_B') DEFPSIST(YES) REPLACE |
On QMGR_B running on hostB.domain.com:9000
| Code: |
DEFINE CHANNEL('CHNL_A_B') CHLTYPE(RQSTR) CONNAME('hostA.domain.com(4500)') REPLACE
DEFINE QLOCAL('QLCL_DATA') DEFPSIST(YES) MAXDEPTH(10000) REPLACE |
What I saw in Production 
On QMGR_A running on hostA.domain.com:4500
| Code: |
DEFINE QLOCAL('QXMT_A_B') DEFPSIST(YES) MAXDEPTH(10000) USAGE(XMIT) REPLACE
DEFINE CHANNEL('CHNL_A_B') CHLTYPE(SDR) CONNAME('hostB.domain.com(9000)') XMITQ('QXMT_A_B') REPLACE |
Note the remote queue manager name:
| Code: |
| DEFINE QREMOTE('QRMT_DATA') RQMNAME('QXMT_A_B') RNAME('QLCL_DATA') XMITQ('QXMT_A_B') DEFPSIST(YES) REPLACE |
On QMGR_B running on hostB.domain.com:9000
| Code: |
DEFINE CHANNEL('CHNL_A_B') CHLTYPE(RQSTR) CONNAME('hostA.domain.com(4500)') REPLACE
DEFINE QLOCAL('QLCL_DATA') DEFPSIST(YES) MAXDEPTH(10000) REPLACE |
And here's another little gem to make this hack work:
| Code: |
| DEFINE QREMOTE('QXMT_A_B') DEFPSIST(YES) REPLACE |
This really works, and the reason for using stated for perpetrating this
evil is as follows:
It simplifies configuring multiple environments without configuring too
much host details into the remote queues.
Questions
1. Why does this even work? Is the remote queue manager name irrelevant?
2. Why does it not work without the remote queue called 'QXMT_A_B'?
3. This is claimed to be IBM endorsed best practice, could that be? 
Last edited by friedl.otto on Wed Feb 06, 2008 6:02 am; edited 2 times in total |
|
| Back to top |
|
 |
| jefflowrey |
| Posted: Wed Feb 06, 2008 5:42 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
You have a typo in your first set of definitions, I think.
I think you meant "RQMNAME('QMGR_B')".
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Feb 06, 2008 6:01 am Post subject: Re: Lengthy ... but interesting channel config question. |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| friedl.otto wrote: |
| 1. Why does this even work? Is the remote queue manager name irrelevant? |
No, but it works because of the queue manager alias
| friedl.otto wrote: |
| 2. Why does it not work without the remote queue called 'QXMT_A_B'? |
Because you need to alias the queue manager name. See 1.
| friedl.otto wrote: |
3. This is claimed to be IBM endorsed best practice, could that be? |
Ask IBM. I've no idea what they endorse for requester channels.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Feb 06, 2008 6:02 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| jefflowrey wrote: |
You have a typo in your first set of definitions, I think.
I think you meant "RQMNAME('QMGR_B')". |
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| friedl.otto |
| Posted: Wed Feb 06, 2008 6:04 am Post subject: |
|
|
Centurion
Joined: 06 Jul 2007
Posts: 116
|
Thanks for the typo alert. FIXED! 
_________________
Here's an idea - don't destroy semaphores unless you're certain of what you're doing! -- Vitor |
|
| Back to top |
|
|
| friedl.otto |
| Posted: Wed Feb 06, 2008 6:08 am Post subject: Re: Lengthy ... but interesting channel config question. |
|
|
Centurion
Joined: 06 Jul 2007
Posts: 116
|
| Vitor wrote: |
No, but it works because of the queue manager alias. |
I don't think I understand ... queue manager alias? All I'm seeing is a
strange 'blank' remote queue definition.
| Vitor wrote: |
Ask IBM. I've no idea what they endorse for requester channels. |
But given the amount of certified greatness around I would have thought
mqseries.net has an idea of the IBM thought processes. 
_________________
Here's an idea - don't destroy semaphores unless you're certain of what you're doing! -- Vitor |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Wed Feb 06, 2008 6:16 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
It's a mistake to assume that a corporation of roughly 300,000 employees has a thought process. *
If that "blank" remote queue definition is on QMGR_B, then it's being used in name resolution. The setup here is the same as a multihop configuration A->B->C, except it's going A->B->B.
*the chips don't work that well yet... ; - )
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Feb 06, 2008 6:20 am Post subject: Re: Lengthy ... but interesting channel config question. |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| friedl.otto wrote: |
| Vitor wrote: |
No, but it works because of the queue manager alias. |
I don't think I understand ... queue manager alias? All I'm seeing is a
strange 'blank' remote queue definition. |
Which we call a "queue manager alias". The Search button is your friend.
| friedl.otto wrote: |
But given the amount of certified greatness around I would have thought
mqseries.net has an idea of the IBM thought processes. |
Many individuals in here do, I personally don't because I don't work for them. It's not impossible someone more authoritative than me may respond, until then I continue recommend that if you want to know why IBM endorse a particular method, you ask IBM.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| friedl.otto |
| Posted: Wed Feb 06, 2008 6:25 am Post subject: |
|
|
Centurion
Joined: 06 Jul 2007
Posts: 116
|
| jefflowrey wrote: |
| It's a mistake to assume that a corporation of roughly 300,000 employees has a thought process. |
He He He ... so the words 'best practice', 'preferred usage pattern' or any
such terminology is pointless.
| jefflowrey wrote: |
| If that "blank" remote queue definition is on QMGR_B, then it's being used in name resolution. The setup here is the same as a multihop configuration A->B->C, except it's going A->B->B. |
I hear what you're saying, but no resolution can come from that object. It
does not reference valid terminal objects.
Here is my super-simplified message traversal:
1. Java application establishes client connection to its queue manager (QMGR_B) on the specified host and specified port.
2. Java application commits an MQ_PUT to the remote queue (QRMT_ONE).
3. The remote queue (QRMT_ONE) is in essence a specialised alias for te transmission queue (QXMT_A_B) which wraps the payload with "XQH" metadata.
4. Sender/Server channel (CHNL_A_B) is notified that it has a pending message on the transmission queue (QXMT_A_B).
5. Sender/Server channel (CHNL_A_B) transmits the "XQH" metadata and payload to the specified host on the specfied port.
6. The Requester channel (CHNL_A_B) receives the "XQH" metadata and payload.
7. The Requester channel (CHNL_A_B) validates the "XQH" metadata.
8. If this is RQMNAME and object RNAME exists try to dereference RNAME.
9. If RNAME could be dereferenced, deliver message.
_________________
Here's an idea - don't destroy semaphores unless you're certain of what you're doing! -- Vitor |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Feb 06, 2008 6:31 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| friedl.otto wrote: |
I hear what you're saying, but no resolution can come from that object. It
does not reference valid terminal objects.
|
Yes it can, yes it does. Look up queue manager aliases as I suggested a few posts back.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| jefflowrey |
| Posted: Wed Feb 06, 2008 6:35 am Post subject: |
|
|
Grand Poobah
Joined: 16 Oct 2002
Posts: 19981
|
| friedl.otto wrote: |
7. The Requester channel (CHNL_A_B) validates the "XQH" metadata.
8. If this is RQMNAME and object RNAME exists try to dereference RNAME.
* At least a couple of missing steps
9. If RNAME could be dereferenced, deliver message. |
There certainly are best practices. Some of them are much more situational than others. To suggest a flavor of this - it is not best practices to use Requester channels, unless it's necessary for some reason.
_________________
I am *not* the model of the modern major general. |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Feb 06, 2008 6:35 am Post subject: |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| friedl.otto wrote: |
| jefflowrey wrote: |
| It's a mistake to assume that a corporation of roughly 300,000 employees has a thought process. |
He He He ... so the words 'best practice', 'preferred usage pattern' or any
such terminology is pointless.
|
It's only valid in so far as it's an accumulation of experience and opinions from 1-n people, where n is a number less than 300,000.
Much like this board really.
Some of the "best practice", like not setting message id and having unique names in clusters, is very much the best way. The advice then slides down through "preferred usage pattern", "suggestion configuration" towards pure opinion.
So you must take what you can from it, as it applies to your situation.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| friedl.otto |
| Posted: Wed Feb 06, 2008 6:35 am Post subject: Re: Lengthy ... but interesting channel config question. |
|
|
Centurion
Joined: 06 Jul 2007
Posts: 116
|
| Vitor wrote: |
| Which we call a "queue manager alias". The Search button is your friend. |
Just to confirm, from this...
* Queue manager alias
* Remote queue resolution
* Aliases on remote queues
... there seems to be a difference between an ALIAS and a QREMOTE. I
suspected as much since a QREMOTE adds "XQH" metadata, and an ALIAS
simply resolves to a QLOCAL. Am I missing something obvious?
_________________
Here's an idea - don't destroy semaphores unless you're certain of what you're doing! -- Vitor |
|
| Back to top |
|
|
| Vitor |
| Posted: Wed Feb 06, 2008 6:40 am Post subject: Re: Lengthy ... but interesting channel config question. |
|
|
Grand High Poobah
Joined: 11 Nov 2005
Posts: 26093
Location: Texas, USA
|
| friedl.otto wrote: |
... there seems to be a difference between an ALIAS and a QREMOTE. I
suspected as much since a QREMOTE adds "XQH" metadata, and an ALIAS
simply resolves to a QLOCAL. Am I missing something obvious? |
No that's about it, though I take some issue with the concept of a QREMOTE "adding" the transmission header. Providing the information for it certainly, but adding it?
See also:
http://www.mqseries.net/phpBB2/viewtopic.php?t=41708&highlight=
for a recent discussion on the difference between a remote queue and an alias queue which may be relevent.
_________________
Honesty is the best policy.
Insanity is the best defence. |
|
| Back to top |
|
|
| friedl.otto |
| Posted: Wed Feb 06, 2008 6:58 am Post subject: Re: Lengthy ... but interesting channel config question. |
|
|
Centurion
Joined: 06 Jul 2007
Posts: 116
|
|
| Back to top |
|
|
|
|
