A written health check of your MQ estate. No agent, no network access.

Consumer payments-settle stalled at 09:42; depth 2,125 of 5,000 and rising 205 msgs/min at export. MAXDEPTH at about 10:05, 14 minutes later.

1. MQSCDISPLAY CONN(*) TYPE(HANDLE) WHERE(OBJNAME EQ PAYMENTS.INBOUND.QL) read-onlyFind the stalled consumer
2. AppRestart the consuming app payments-settle on payapp03 change, run by your teamApplication change, made by your team
3. MQSCDISPLAY QLOCAL(PAYMENTS.INBOUND.QL) CURDEPTH read-onlyConfirm depth is falling

The weekday-morning backlog for the 6th time in 30 days, steeper each time: 31,400 at export, rising 138 msgs/min, MAXDEPTH at about 12:05. Permanent fix: Run two orders-dispatch consumers all day (the queue already allows shared input), then confirm a weekday morning passes without the climb.

1. MQSCDISPLAY QSTATUS(ORDERS.OUTBOUND.QL) TYPE(QUEUE) CURDEPTH IPPROCS MSGAGE read-onlyConfirm one consumer and the climb
2. AppStart a second orders-dispatch instance change, run by your teamApplication change, made by your team (the queue allows shared input)
3. MQSCDISPLAY QSTATUS(ORDERS.OUTBOUND.QL) TYPE(QUEUE) CURDEPTH IPPROCS read-onlyExpect IPPROCS(2) and depth falling

CF structure 54.3% used and projected full at about 20:40, before the 22:00 drain. Puts to SETTLE.SHARED.QL would then fail. Longer term: CSQI004I recommends indexing SETTLE.SHARED.QL by CORRELID so the nightly drain stops scanning past messages; plan that change with your z/OS team.

1. MQSCDISPLAY CFSTATUS(SETLSTR1) TYPE(SUMMARY) read-onlyConfirm usage and entries in use
2. z/OSD XCF,STR,STRNAME=PSG1SETLSTR1 read-onlyCheck current and maximum size in the CFRM policy
3. z/OSSETXCF START,ALTER,STRNAME=PSG1SETLSTR1,SIZE=6291456 change, run by your teamExpand to 6 GB if the CFRM policy allows; run by your z/OS team
4. MQSCDISPLAY CFSTATUS(SETLSTR1) TYPE(SUMMARY) read-onlyVerify the new size and usage

Stopped 3 times in 30 days when its security exit ended abnormally (last on 23 Oct, 03:14 to 03:22). Until now: START CHANNEL(BANK.TO.CLEARING.SDR) after each stop. Permanent fix: Raise the clrsecx failure with its vendor and move to a fixed exit level; until then the stops will recur.

1. MQSCDISPLAY CHSTATUS(BANK.TO.CLEARING.SDR) read-onlyConfirm STATUS(STOPPED) and the substate

Rejected at connect with reason 2035 on 25 Sep and 23 Oct, both on credential-rotation nights. Permanent fix: Add the settlement-batch secret to the rotation runbook so it is deployed before the 02:45 batch.

1. MQSCDISPLAY QMGR CONNAUTH read-onlyConfirm which AUTHINFO object applies
2. MQSCDISPLAY AUTHINFO(SYSTEM.DEFAULT.AUTHINFO.IDPWOS) CHCKCLNT read-onlyConfirm CHCKCLNT(REQUIRED); do not relax it

No dead-letter queue appears in this queue manager's export. Confirm DEADQ is set, so undeliverable messages are kept rather than blocking a channel.

1. MQSCDISPLAY QMGR DEADQ read-onlyExpect the name of a defined dead-letter queue

6 times: 30 Sep, 7 Oct, 14 Oct, 16 Oct, 22 Oct, 23 Oct (fingerprint FP-7Q2K).

Until now: Restart orders-dispatch, or start a second instance by hand, each morning it happens.

Permanent fix: Run two orders-dispatch consumers all day (the queue already allows shared input), then confirm a weekday morning passes without the climb.

3 times: 2 Oct, 11 Oct, 23 Oct (fingerprint FP-3MX9).

Until now: START CHANNEL(BANK.TO.CLEARING.SDR) after each stop.

Permanent fix: Raise the clrsecx failure with its vendor and move to a fixed exit level; until then the stops will recur.

2 times: 25 Sep, 23 Oct (fingerprint FP-8D1C).

Until now: Deploy the new credential by hand after the batch fails.

Permanent fix: Add the settlement-batch secret to the rotation runbook so it is deployed before the 02:45 batch.