IBM MQ V9.1 System Administrator: Troubleshooting - IBM MQ V9.1 System Administrator

Troubleshooting - what the exam tests

This domain tests diagnosis: reading AMQ error logs and FFST files, interpreting reason codes, checking channel and queue status, handling the dead-letter queue, and using trace only when logs are insufficient.

Core rules

• Logs: newest errors are in /var/mqm/qmgrs/QM1/errors/AMQERR01.LOG; AMQERR02.LOG and 03 are older, not newer. Global errors and .FDC FFST files are under /var/mqm/errors.
• Reason codes: 2035 NOT_AUTHORIZED, 2058 Q_MGR_NAME_ERROR, 2059 Q_MGR_NOT_AVAILABLE, 2053 Q_FULL, 2033 NO_MSG_AVAILABLE, 2085 UNKNOWN_OBJECT_NAME, 2009 CONNECTION_BROKEN, 2018 CONNECTION_HANDLE_ERROR. Match the code to the layer before changing anything.
• Channels: DISPLAY CHSTATUS(ch) ALL shows RUNNING, RETRYING, STOPPED, substate, XQMSGSA, last message time. PING CHANNEL(ch) tests the link. RESOLVE CHANNEL(ch) ACTION(COMMIT|BACK) settles in-doubt; RESET CHANNEL(ch) resets sequence numbers and can discard messages, so use it only after comparing both ends.
• DLQ: undeliverable messages land on SYSTEM.DEAD.LETTER.QUEUE with an MQDLH reason. Browse before destructive get; runmqdlq applies a rules table to retry or forward.
• Trace: strmqtrc -m QM1, endmqtrc -m QM1, dspmqtrc for formatted output; trace is for defects, not ordinary 2035s.

Worked example

A sender channel is RETRYING and XQMSGSA grows. DISPLAY CHSTATUS shows a connection error; PING CHANNEL fails. Reasoning: check the remote listener with DISPLAY LSSTATUS(TCP), confirm CONNAME host/port and firewall, and confirm the receiver channel exists with the same name. If the listener is running and the name matches, examine the remote queue manager's AMQERR01 for a 2059 or TLS error. Do not RESET CHANNEL yet: sequence numbers are not the problem while the TCP connection cannot be established.

Traps

Treating 2035 as a network fault wastes time; it is an authority failure. Looking only at the local queue manager hides receiver-side errors. RESET CHANNEL to "fix" a stopped channel can lose in-flight messages. Ignoring a growing DLQ means production messages are being discarded. Reading AMQERR03 first finds the oldest, not newest, error.