During investigation held at 16/9/25  on ΑΙΧ ΜQ host -where production (Credit Risk) Queue Manager resides- it has been identified that hundreds of "GHOSTS" durable topic subscriptions have been released by (Credit Risk) producer apps  which are stored on hundreds System Managed Durable Local Queues hosted on Queue Manager, causing storage space on path:  /var/mqm/qmgrs/QMGR/queues to be exhausted..

Given that each managed subscription is associated to its own system managed durable queue, the problem arises even if producer application is not connected and not consuming messages from the managed subscription queue, since messages will continue to be delivered to that queue.. 

FYI, the same problem occurs on QA environment with lower numbers due to lower frequency of flows

Temporary workaround is to provide  more storage space periodically (same problem / same symptom occurred also at 10/2/25)

Evidences:

"GHOSTS" durable topic subscriptions timestamps (sample)

Abstract of system managed durable queues (approx. no: 1400!!) where durable subscriptions maintained, hosted on QM

My questions are the following:

1)  is it advisable to delete durable subscriptions &  with older date (i.e. 1 month old) & System Managed Durable Local Queues without msgs (current depth=0) without impacting the stability of topics publish/subscription flows?

2) Is it advisable for App Vendor to examine the option of producing non-durable published subscriptions that are not delivered if not consumed? 

Looking for your advise, on the above questions or any other remediation activity from MQ side (apart from app)

P.S.

App owners have just raised a ticket to app vendor expecting his response.

Thanks & Rgds,

Nick.