| Author |
Message
|
| Jakob_CPH |
 Posted: Thu Aug 11, 2005 3:19 am Post subject: No repository |
 |
|
Apprentice
Joined: 28 Jan 2005
Posts: 26
|
Hi,
I have a problem with a queue manager IQP4 on z/OS in a EDI cluster where queue manager A and B is repository managers. My CHIN says
| Quote: |
| IQP4 CSQXREPO No repositories for cluster EDI |
There is a cluster channel saying A is repository manager
| Quote: |
Cluster queue manager name : A
Channel name . . . . . . . : TO.A
Channel definition type . . : CLUSSDRB
Cluster name . . . . . . . : EDI
Disposition . . . . . . . . : - IQP4
Queue manager identifier . : A_2002-06-07_15
Repository management . . . : Y Y=Yes, N=No
Suspended . . . . . . . . . : N Y=Yes, N=No |
After CLUSTER REFRESH the CHIN still think there is no repository for EDI cluster.
Why is that so?
Is it a problem?
Thanks
Jakob Hansen |
|
| Back to top |
|
 |
| srinivasACN |
| Posted: Thu Aug 11, 2005 7:05 am Post subject: |
|
|
Apprentice
Joined: 08 Aug 2005
Posts: 43
|
Do you have a receiver channel on IPQ4 coming from QM A?
When setting up a cluster, every QM (which are partial repositories) need to have a cluster sender and receiver channel. Sender going to the repository and a cluster reciever.
Other than that, nothing else is necessary and after doing all this a REFRESH CLUSTER should take care of it. |
|
| Back to top |
|
|
| srinivasACN |
| Posted: Thu Aug 11, 2005 7:08 am Post subject: |
|
|
Apprentice
Joined: 08 Aug 2005
Posts: 43
|
| srinivasACN wrote: |
| Do you have a receiver channel on IPQ4 coming from QM A? |
I correct myself, not coming from A but a general cluster receiver channel on IPQ4 |
|
| Back to top |
|
|
| Jakob_CPH |
| Posted: Thu Aug 11, 2005 8:06 am Post subject: |
|
|
Apprentice
Joined: 28 Jan 2005
Posts: 26
|
Yes there is a a cluster receiver channel for the EDI cluster on qmgr IQP4.
To write something about the problems we had originally here is some writing I made regarding the problem
| Quote: |
Repository Manager Resilience
Friday the 29th of August we had more than 2hours down time in production for a business critical application SAAC API, because the repository manager on the Z/OS queue manager IQP4 was down.
The problem was that when the broker on IQP4 tried to send to cluster queues it received back a MQ error code 2269. Therefore messages were not send to BROKER.ACS.DELIVERY.REQUEST on the SAAC API queue manager SAACPAQM.
The reason the repository manager was down was an invalid message in the SYSTEM.CLUSTER.COMMAND.QUEUE on IQP4 which the repository manager could not execute. Therefore it stopped.
The problem was solved when the command queue on IQP4 was flushed and the queue manager was restarted.
The course of the invalid message
The invalid message in the command queue properly came from a EDIPRD queue manager with a badly configured cluster channel.
There are four queue managers and two clusters:
- The queue managers EDIPRD and EDIPRDB are repository manger for the cluster EDIPRD.
- The queue managers IQP4 and IQP5 are repository managers for cluster BRK_EDI.
The manually defined channels on EDIPRD are
DEFINE CHANNEL ('TO.IQP4.BRK_EDI') CHLTYPE(CLUSSDR) +
* ALTDATE (2005-07-27) +
* ALTTIME (17.07.54) +
TRPTYPE(TCP) +
BATCHINT(0) +
BATCHHB(0) +
BATCHSZ(50) +
CLUSTER(' ') +
CLUSNL('EDINLPRD') +
DEFINE CHANNEL ('TO.EDIPRD.BRK_EDI') CHLTYPE(CLUSRCVR) +
* ALTDATE (2005-07-2 +
* ALTTIME (16.47.3 +
TRPTYPE(TCP) +
BATCHINT(0) +
BATCHHB(0) +
BATCHSZ(50) +
CLUSTER('EDIPRD') +
CLUSNL(' ') +
The namelist EDINLPRD has the content (EDI_BRK and EDIPRD)
The cluster channels manually defined on IQP4 are
TO.IQP4.BRK_EDI (cluster receiver) member of cluster EDI_BRK :
TO.IQP5.BRK_EDI (cluster sender) member of cluster EDI_BRK
But the automatically defined EDI cluster channels are
Channel name Connection name State
Type Cluster name Suspended
Cluster queue manager name Disposition
TO.EDIPRD EDIPRDMQ.CRB.APMOLLER.NET(1414) INACT
CLUSSDRA EDIPRD N
EDIPRD - IQP4
TO.EDIPRD.BRK_EDI EDIPRDMQ.CRB.APMOLLER.NET(1414) INACT
CLUSSDRA BRK_EDI N
EDIPRD - IQP4
TO.EDIPRD.IQP7 BROKER1.APMOLLER.NET(51421) INACT
CLUSSDRA EDIPRD N
IQP7 - IQP4
TO.EDIPRD.SAFCPAQM cptmqs1.crb.apmoller.net(1422) INACT
CLUSSDRA EDIPRD N
SAFCPAQM - IQP4
TO.EDIPRDB MQ-HOST.BRY.DK.EUR.CRB.APMOLLER.NET(1415) INACT
CLUSSDRA EDIPRD N
EDIPRDB - IQP4
******** End of list ********
Please note the four channels member of the EDIPRD cluster even though IQP4 not has manually defined channels member of EDIPRD.
We think the bad message in the command queue maybe came from the EDIPRD queue manager.
Conclusion
The main problem is that the repository manager goes down because there is a message it can not process. This is all overall risk for the mq clustering on the queue manager. It is not acceptable that messages can not be put on a cluster queue because there is an invalid message on the command queue.
|
|
|
| Back to top |
|
|
| fjb_saper |
| Posted: Thu Aug 11, 2005 10:50 am Post subject: |
|
|
Grand High Poobah
Joined: 18 Nov 2003
Posts: 20756
Location: LI,NY
|
On your clusterreceiver channel did you not define the conname or did you just not show it ?
 |
|
| Back to top |
|
|
| Jakob_CPH |
| Posted: Thu Aug 11, 2005 10:49 pm Post subject: |
|
|
Apprentice
Joined: 28 Jan 2005
Posts: 26
|
| All the settings are now correct so I suspect there is a wrong message in the repository queue(?) |
|
| Back to top |
|
|
|
|
