Closed rjseibert closed 3 years ago
I should note that this may be related to queues with usage=XMITQ being interpreted as usage=NORMAL queues.
This is possibly related to https://github.com/ibm-messaging/mq-golang/issues/133.
It does sound the same as #133. It looks like you've set the monitoredQueuePatterns
parameter - possibly set to '*'. How many queues do you have defined that match that pattern? And since it's z/OS, is the qmgr in a QSG?
We have between 600 and 700 queues. Some are in QSGs and some are not.
Thanks for that. The numbers combined with the stack info have given me a couple of theories. One is around the resizing of buffers.
In vendor.../mqmetric/discover.go, round about line 860, there's a line bufsize:=32768
. If you change that to 327680, it might now work for you. If it does work, that wouldn't be a proper fix, and it's not my only theory, but at least it would prove something.
@ibmmqmet, your suggestion did indeed work. I opened a PR for increasing the bufsize, but I do understand that it is not a "proper" fix.
Properly fixed in v5.2.3
Please include the following information in your ticket.
- Version information for MQ, mq-golang, Go compiler
I am connecting to a z/OS queue manager.
- A small code sample that demonstrates the issue.
I found this snippet of code: https://github.com/ibm-messaging/mq-golang/blob/master/ibmmq/mqiPCF.go#L298
I find the comment particularly humorous.