Most IBM MQ tutorials on this site help you configure queues, open channels, secure listeners, and design integration patterns—the skills you use weekly as an administrator or integrator. The advanced internals track goes under the hood: what the queue manager writes to disk when a message is persistent, how the log serializes changes, which processes move bytes on channels, and why a restart after power loss either recovers cleanly or leaves the repository in need of IBM support. You do not need this material to pass your first put/get lab, but when production shows unexplained message loss, hours-long restart, or channel stalls with no obvious network fault, internals knowledge turns guesswork into hypothesis and evidence. This page is the roadmap: prerequisites, recommended order, lab safety, relationship to trace and FFST files, and how distributed Linux queue managers compare to z/OS at a high level before you open the dedicated tutorials on logging, persistence, channel agents, repository manager, pub/sub engine, buffers, locking, and recovery.
| Order | Topic | Why this order |
|---|---|---|
| 1 | Logging internals | Logs underpin recovery; read first |
| 2 | Message persistence internals | How messages land on disk |
| 3 | Channel agents | Movement between queue managers |
| 4 | Repository manager internals | Object definitions storage |
| 5 | Pub/sub engine internals | Topic routing machinery |
| 6 | Buffer management | Memory versus disk tuning |
| 7 | Locking | Contention and deadlock symptoms |
| 8 | Recovery processing | Ties log + persistence together |
Learning queues and channels is driver's education. The internals track is mechanic school—you learn what happens inside the engine when you turn the key so you can fix it on the highway, not just drive to the store.
Never experiment with log deletion, forced mqrmirecovery flags, or repository edits on production without IBM guidance and backup. Use disposable queue managers in VMs or containers. Copy FDC and logs to a case folder before cleanup. One reproducible test beats ten random changes.
1234567# Illustrative diagnostic commands — lab only dspmqver dspmq -m QM1 runmqsc QM1 << EOF DISPLAY QMSTATUS ALL EOF # strmqtrc -t detail -o /tmp/mqtrace (stop with endmqtrc per doc)
Distributed queue managers use files under /var/mqm or /mnt/mqm with log and qm repositories in layouts documented by IBM. z/OS uses page sets, BSDS analogs in concept, and coupling facility for shared queues. Internal module names differ; mental models of write-ahead logging and channel movers still transfer.
If application MQRC points to authorization or wrong queue name, fix config before tracing internals. Escalate to IBM support with FFST, AMQ logs, and trace when corruption or unexplained persistence loss appears. Internals study does not replace support for defect suspicion.
This track is a map of lessons about how the mail-sorting machine works inside, for people who fix the machine when mail gets stuck—not just people who drop letters in the slot.
List which prerequisite tutorials you have completed.
Build a lab QM document with backup and trace commands allowed.
Pick one production incident and note which internals topic would explain it.
1. Internals track is for:
2. Before internals learn:
3. strmqtrc is:
4. z/OS MQ internals: