Connection failures while using MQ adpater to put the data.

[snehareddy]

Hello ALL,

We are getting connection failures while the maps are using the MQ adapter to put the data on IBM websphere MQ.

The following the return code we are getting :

Adapter Type: IBM WebSphere MQ (S
Resource Name: -qmn XZCSR1 -qn CESAR_AUDIT
Adapter Message: Connection failed
Return Code: -13
Data Size: 72

We are using mercator version 6.7 and version of the IBM websphere MQ is
WebSphere MQ
Version: 530.4 CSD04
CMVC level: p530-04-030617

These errors have been observed recently after there is increase in the number of maps which are trying to connect to the queue manager using MQ adapter.

The errors are stopping after the MQ was restarted and also eventservers were restarted.

Has anyone faced this problems earlier and if yes what was the causes and possible solution.

A quick help in this regard would be really great.

Thanks in advance.

Sneha

 

[OllyC]

try setting this environment variable for the TX user before starting the event server:

export EXTSHM=ON

(assuming your running on UNIX)

if this doesn't work, whats the full adapter string?

 

[snehareddy]

Hello,

Thanks for the information,can you please inform what does setting EXTSHM=ON refers to and the complete adapter string is -omMQS2 '-qmn XZCSR1 -qn CESAR_AUDIT'.

Thanks and Regards,
Sneha

 

[OllyC]

EXTSHM is an environment variable, it turns extended shared memory on.

Did it fix the problem?

 

[snehareddy]

Hello,

Thanks for the information,this is problem is occuring on our production environment and it has been observed after there is a steep increase in the volume of messages that are put on the IBM websphere queue using MQ adapter and also we are not able to replicate the problem on our test environments and hence would like to know the following before we go ahead and apply the change suggested above.

1.Does this mean that as number of connections that are made by the MQ adapter increase, the connection failures occur due to inadequate memory.

2. By extending the shared memory how will it affect the MQ adapter usuage interms of volumes.

Thanks in advance!!!

Regards,
Sneha

 

[JEMIAZ]

We have had this problem a long time in the past and it was caused because we had miss configured the queue manager (according to the wtx manual).

The QM maxhands setting should be >1024 while the default is 256 (I think). If the QM is busy it may not be able to keep track of all the threads in a timely manner.

From runmqsc XZCSR1 on the server type display qmgr maxhands to get the current value.

the alternative is to check that there are not 2 event server processes running against the same QM.
Tim

 

[snehareddy]

Hello,

Thanks for the response,we had checked the value of Max threads is 256,we never had the MQ connection failures earlier with this value,now there is a steep increase in number messages that are being put on the queue manager(200,000 messages/day) does increase in volume need this change on the queue manager,if yes we will go ahead and change the same.

Can you please clarify second point check that there are not 2 event server processes running against the same QM.

Also can you please my earlier questions


1.Does this mean that as number of connections that are made by the MQ adapter increase, the connection failures occur due to inadequate memory.

2. By extending the shared memory how will it affect the MQ adapter usuage interms of volumes.

Thanks in advance!!!

Regards,
Sneha

 

[JEMIAZ]

On Windows you can get 2 Event Servers running using the same .msl's and you get double triggering. With MQ one of the maps will work ok and the other will fail often siting MQ connection error -13 and map error 12 (source not available)

I'm not sure what the requirement is for increasing from 256. Its probably a combination of the number of mq listeners you have on different queues, whether the maps are multi-threaded and whether the data arrives in bursts or a steady stream.

Certainly the manual will say its too low a value.

Have you got the latest MQ adapter related patches from support?

You may also want to uplift the MQ patch level. Fix pack 13 is available but IBM will only to support up to 11 I think. MQ 6.0 is not supported on release 6.7

 

[OllyC]

>1.Does this mean that as number of connections that are made by the MQ adapter increase, the connection failures occur due to inadequate memory.

We don't know your server details, OS details, details of your maps and data + you're running a very old version of Mercator - sometimes you just need to try stuff out to see if it works.

2. By extending the shared memory how will it affect the MQ adapter usuage interms of volumes.

as above, try it out. If it fixes the problem the problem then great.

See

http://middleware.its.state.nc.us/middleware/Documentation/en_US/htm/csqzal06/csqzal064e.htm