Join our Newsletter — 33% off our NHI Course
Home› Guides› IBM MQ Security Guide: Channel Authentication, TLS and…
Guide Application Security

IBM MQ Security Guide: Channel Authentication, TLS and Queue Manager Access

PDF for newsletter subscribers · free to read online

← All guides
By NHI Mgmt Group, edited by Lalit Choda Updated 14 November 2026 16 min read
On this page

An IBM MQ queue manager is only as secure as the user ID each channel runs under, and that ID is decided by a chain of settings most teams never write down. Every application, channel and agent that connects to a queue manager is a non-human identity, and the customer controls how each one is identified, authenticated and authorised: channel authentication records (CHLAUTH), connection authentication (CONNAUTH), the MCAUSER on each server-connection channel, object authority manager (OAM) records, TLS on channels and, where messages must be protected at rest, Advanced Message Security (AMS). This guide explains how those layers fit together on IBM MQ for Multiplatforms 9.4, with z/OS differences noted, and gives a production baseline, a rollout plan and the events to monitor. It is for MQ administrators, integration engineers and security architects.

This is an independent guide from NHI Mgmt Group, based on IBM's public IBM MQ documentation as of November 2026. It is not sponsored or reviewed by IBM. Product features change, so confirm settings against IBM's current documentation.

Key takeaways

  • Decide the MCAUSER, never inherit it. A blank MCAUSER on a server-connection channel lets a client connect with the authority of whatever user ID it supplies, so set a low-privilege MCAUSER or map one with a CHLAUTH rule on every SVRCONN channel.
  • CHLAUTH decides who reaches the queue manager; OAM decides what they can do. Use CHLAUTH to block, map and pin identities by address, certificate DN, queue manager name or asserted user, then grant narrow OAM authorities to the mapped IDs.
  • Turn on credential checks for clients. New queue managers point CONNAUTH at SYSTEM.DEFAULT.AUTHINFO.IDPWOS with CHCKCLNT(REQDADM); move application channels to REQUIRED, or to mutual TLS with SSLCAUTH(REQUIRED), so no client connects on an asserted ID alone.
  • Use TLS on every channel that crosses a network. Set SSLCIPH to a TLS 1.3 or TLS 1.2 CipherSpec on both ends, filter peers with SSLPEER or SSLPEERMAP rules, and keep deprecated CipherSpecs disabled, which is the default.
  • Turn on the events you need to see misuse. Authority events (AUTHOREV) record refused requests on Multiplatforms and, from 9.4.3, can record successful ones too; add channel, configuration and command events and ship them to your SIEM.

How identity and access work in IBM MQ

A client connection passes through four decisions, in this order. Getting any one wrong usually means a connection runs with more authority than intended.

LayerSettingWhat it decidesKey objects and commands
1. Who may connectCHLAUTH recordsBlock or allow by IP address, user ID, TLS DN or remote queue manager, and map the connection to a run-time user IDALTER QMGR CHLAUTH(ENABLED), SET CHLAUTH, DISPLAY CHLAUTH
2. Are the credentials validCONNAUTH and AUTHINFOWhether a user ID and password must be supplied and checked against the OS or LDAPAUTHINFO type IDPWOS or IDPWLDAP, CHCKCLNT, CHCKLOCL, ADOPTCTX
3. Which ID the channel runs asMCAUSER resolutionThe user ID used for every MQI call the server-connection channel makesSecurity exit, CHLAUTH mapping, channel MCAUSER, asserted client ID
4. What that ID may doOAM authority recordsConnect, put, get, browse, inquire, set context and administer named objectssetmqaut, dspmqaut, dmpmqaut

IBM documents the precedence for the server-connection user ID: a user ID set by a security exit is used first (unless a BLOCKUSER rule blocks it), then one set by a CHLAUTH rule, then the MCAUSER on the channel, then the user ID asserted by the client, and finally the ID that started the channel. The default SYSTEM.DEF.SVRCONN channel has a blank MCAUSER, and IBM recommends changing it to a user ID with no access to IBM MQ objects. On z/OS the same model applies, but authority checks are made by RACF (or another external security manager) against classes such as MQCONN and MQQUEUE; see the Mainframe and COBOL Application Security Guide and the Mainframe Identity and RACF Security Guide.

Settings to review on day one

  • Blank MCAUSER on SVRCONN channels: set SYSTEM.DEF.SVRCONN and any unused SYSTEM.* channels to a user ID with no MQ authority, and block them with CHLAUTH.
  • Privileged user IDs over client channels: keep a BLOCKUSER rule that blocks *MQADMIN on application channels. *MQADMIN means members of the mqm group on AIX and Linux, the mqm and Administrators groups and SYSTEM on Windows, and the address space user IDs on z/OS.
  • CONNAUTH on migrated queue managers: queue managers migrated from releases before 8.0 have a blank CONNAUTH attribute, so no credential checking happens until you set it.
  • Channels with SSLCIPH blank: a blank SSLCIPH means TLS is off on that channel.

Channel authentication rules

CHLAUTH records are matched against each inbound channel. Rules apply only when the queue manager attribute CHLAUTH is ENABLED. The record types:

TYPEMatches onTypical use
BLOCKADDRIP address, range or patternRefuse known-bad or out-of-scope networks for all channels; blocked connections are held for 30 seconds before the error is returned
BLOCKUSERUser ID list, no wildcardsBlock *MQADMIN and named IDs on client channels (not message channels)
ADDRESSMAPIP address, range or patternMap an application subnet to a dedicated MCAUSER, or block with USERSRC(NOACCESS) on named channels
USERMAPClient-asserted user IDMap an asserted ID to a run-time ID; best combined with credential checks, because asserted IDs are not trustworthy on their own
SSLPEERMAPTLS certificate DN, wildcards allowedMap each application certificate to its own MCAUSER: the strongest pattern for application identity
QMGRMAPRemote queue manager namePin which queue managers may use a receiver or cluster channel and with which ID

USERSRC controls where the run-time ID comes from: MAP (the default) uses the MCAUSER in the record, NOACCESS ends the channel immediately, and CHANNEL uses the flowed ID or the channel's MCAUSER. On mapping records, CHCKCLNT can tighten the queue manager default per rule: ASQMGR follows the queue manager setting, REQUIRED demands valid credentials, and REQDADM demands them only for privileged IDs (not valid on z/OS).

Testing safely: WARN(YES) on a blocking record lets a connection that would be blocked through and can record a channel event, so you can trial a new rule set before enforcing it. Message AMQ9787 for warnings also needs ChlauthIssueWarn=y in the Channels stanza of qm.ini.

Connection authentication and MCAUSER design

CONNAUTH points at an AUTHINFO object of type IDPWOS (local operating system) or IDPWLDAP (LDAP). New queue managers set it to SYSTEM.DEFAULT.AUTHINFO.IDPWOS with CHCKCLNT(REQDADM), so privileged users must supply valid credentials while other users are treated as OPTIONAL: any credentials supplied must be valid, but none are required. The values:

CHCKCLNT / CHCKLOCLEffectWhen to use
NONECredentials are not checkedOnly where TLS client certificates plus SSLPEERMAP rules carry the identity
OPTIONALSupplied credentials must be valid, but supplying them is optionalTransition state while applications are updated
REQDADMPrivileged IDs must authenticate; others as OPTIONALDefault starting point on Multiplatforms
REQUIREDEvery application must supply valid credentialsTarget for client connections that do not use mutual TLS

ADOPTCTX(YES) makes the authenticated user ID the context for authorisation checks; with NO the credentials are only checked and the application's own ID is used afterwards. Use YES so that the identity you authenticated is the identity OAM checks. After any change, run REFRESH SECURITY TYPE(CONNAUTH): the queue manager does not pick up the change until you do.

Choosing between passwords and certificates: passwords over CONNAUTH are simpler to start with but are credentials that sit in application configuration and must be rotated; see the Secrets Management Guide. Mutual TLS with SSLCAUTH(REQUIRED) and an SSLPEERMAP rule per application avoids a shared secret and gives each application a distinct, revocable identity, at the cost of certificate lifecycle work; see the mTLS Guide. Large estates usually end up with certificates for applications and LDAP-backed passwords for people.

MCAUSER design: IBM recommends a dedicated MCAUSER with limited authority, separate IDs for client groups that need different access, and not relying on client-supplied IDs. In practice that means one run-time ID per application, not one per channel or one for everything, so OAM records and events point at a single owner. Treat these IDs as service accounts: the Service Account Security Guide covers ownership and review.

Object authorities with the OAM

On Multiplatforms the OAM is the default authorisation service and is active by default. It grants authorities to principals and groups (on AIX and Linux, to groups), and is administered with setmqaut, dspmqaut and dmpmqaut or the equivalent MQSC authority record commands. Practical rules:

  • Grant to dedicated groups, one per application: connect and inquire on the queue manager; put on the queues it sends to; get and browse on the queues it reads; nothing generic.
  • Treat context and alternate-user authorities as privileged: authorities that let an application set identity context or act as another user let it impersonate other applications in message headers. Grant them only to components that genuinely need them, such as bridges.
  • Keep administrative authority off application IDs: no application ID should be in the mqm group or hold change, delete or control authority on queue manager objects.
  • Use generic profiles carefully: a profile on APP1.** is tidy; a profile on ** is an administrator by another name.
  • Dump and diff: dmpmqaut output stored in version control gives you a baseline to review and an easy way to spot drift.

TLS on channels

TLS is configured per channel with ALTER CHANNEL. Set SSLCIPH to a CipherSpec valid on the platform; both ends of a channel must use compatible CipherSpecs or the channel will not start, and a blank SSLCIPH turns TLS off. Key settings:

  • SSLCIPH: choose a TLS 1.3 CipherSpec, or a TLS 1.2 one where a partner cannot yet do 1.3. Deprecated CipherSpecs cannot be set on a channel by default (the command fails with AMQ8242), and IBM MQ 9.4.4 removed a set of CipherSpecs from Multiplatforms that cannot be re-enabled.
  • Keep weak suites off: the AMQ_SSL_WEAK_CIPHER_ENABLE environment variable and AllowWeakCipherSpec in the SSL stanza of qm.ini re-enable deprecated suites, and on z/OS dummy DD statements such as CSQXWEAK do the same. Alert on any of them appearing in configuration.
  • SSLCAUTH: REQUIRED rejects any client that does not present a valid certificate; OPTIONAL checks a certificate only if one is sent. Use REQUIRED on application SVRCONN channels that use mutual TLS.
  • SSLPEER and SSLPEERMAP: SSLPEER on the channel restricts accepted certificates by DN attributes (CN, O, OU and so on, with a leading or trailing wildcard); SSLPEERMAP CHLAUTH rules go further and map each DN to its own MCAUSER. Quote values to preserve case.

Certificates for queue managers and applications are machine identities with expiry dates; an expired queue manager certificate stops every TLS channel at once. Track them with the rest of your certificate estate; see the Machine Identity, PKI and Certificate Lifecycle Guide.

Protecting messages with Advanced Message Security

TLS protects a message on the wire between two channel ends; it does not protect it while it sits on a queue or on disk, or from an administrator who can browse the queue. AMS adds protection to the message itself using public key cryptography, defined per queue with setmqspl (the queue manager must be running, and the policy name matches the queue name). The three qualities of protection:

Quality of protectionsetmqspl settingsProtects againstNote
IntegritySignature algorithm (-s), no encryptionTampering and unauthorised senders (with -a authorised signers)Key reuse must be 0
PrivacySignature and encryption (-e) with recipients (-r)Disclosure and tampering, end to endKey reuse must be 0
ConfidentialityEncryption with recipients, signature NONEDisclosure, with lower overheadKey reuse (-c) allowed

Choose SHA256 or stronger for signatures and AES128 or AES256 for encryption; the command still accepts older algorithms (MD5, SHA1, RC2, DES, 3DES), several of which IBM marks as deprecated. The toleration flag -t 1 lets applications read unprotected messages during a staged rollout; set it back to 0 once every producer is protected. Use AMS for queues carrying payment, personal or credential data where the threat includes insiders and stored messages, not only network sniffing.

Administrative access and the web console

  • Remote MQSC and PCF administration: administration over SYSTEM.ADMIN.SVRCONN or a custom admin channel should require TLS with SSLCAUTH(REQUIRED), an SSLPEERMAP rule per administrator or tool, and CHCKCLNT(REQUIRED). Restrict it with ADDRESSMAP rules to management subnets.
  • mqweb (IBM MQ Console and REST API): users and roles are defined in mqwebuser.xml against a basic, LDAP, local OS or (on z/OS) SAF registry; the registry is empty by default. Use LDAP or SAF rather than the basic registry in production, use HTTPS (basic authentication is only secure over a secure connection), and prefer client certificates or short LTPA token expiry. Grant administrator roles to named people and read-only roles to monitoring tools.
  • The mqm group: membership is full control of every queue manager on the host. Review it like any privileged group, and keep application and monitoring IDs out of it.

Automation that administers MQ (deployment pipelines, monitoring agents, managed file transfer agents) holds MQ credentials or certificates. Inventory them as non-human identities, give each its own ID and CHLAUTH mapping, and rotate their credentials. The Managed File Transfer (MFT) Security Guide covers the file transfer side.

Logging and detection

SourceEnable withWhat to alert on
Authority events (SYSTEM.ADMIN.QMGR.EVENT)ALTER QMGR AUTHOREV(ENABLED); from 9.4.3 AUTHEVSC sets whether successful checks are also reportedBursts of Not Authorized events from one ID or address; any refused request against administrative objects
Channel and TLS eventsChannel event switches (CHLEV, SSLEV) on the queue managerChannels blocked by CHLAUTH, TLS handshake failures, unexpected new channel instances
Configuration events (SYSTEM.ADMIN.CONFIG.EVENT)CONFIGEVChanges to channels, CHLAUTH records and AUTHINFO objects outside change windows
Command eventsCMDEVSecurity commands such as SET CHLAUTH and authority changes from unexpected users
Queue manager error logsOn by default (AMQERR files)AMQ9776 (channel blocked by user ID), AMQ9787 (warning mode), repeated authentication failures

Measure that the controls hold, quarterly:

  • SVRCONN channels with a blank MCAUSER and no mapping CHLAUTH rule (target 0).
  • Network-facing channels with SSLCIPH blank (target 0).
  • Client connections authenticated by certificate or REQUIRED credentials, as a share of all client connections (target 100% for production).
  • Application IDs holding authority on ** profiles or in the mqm group (target 0).
  • Queue manager and application certificates expiring within 30 days without a renewal ticket (target 0).
  • Not Authorized events per week by application, with a named owner for each recurring source.

Rolling out a production baseline

  1. Inventory: list every queue manager, channel, SSLCIPH value, MCAUSER, CHLAUTH record (DISPLAY CHLAUTH(*)) and OAM dump. Map each SVRCONN channel to the applications that use it.
  2. Enable events first: switch on authority, channel, configuration and command events so you can see what each later change breaks.
  3. Add CHLAUTH rules in warning mode: create the mapping and blocking rules with WARN(YES), run for one business cycle, and fix every application that would be blocked.
  4. Enforce in rings: development, then test, then one production queue manager, then the rest. Remove WARN(YES) ring by ring.
  5. Raise credential checks: move CHCKCLNT to REQUIRED, or SSLCAUTH to REQUIRED with SSLPEERMAP, per channel once each application supplies credentials or a certificate; run REFRESH SECURITY TYPE(CONNAUTH).
  6. Tighten OAM: replace broad grants with per-application groups and narrow profiles; keep the previous dmpmqaut output.
  7. Add AMS where data needs it: start with toleration on, then turn it off once every producer signs or encrypts.

Rollback: each step is reversible. Re-add WARN(YES) to a rule, return CHCKCLNT to its previous value and refresh, or replay the saved dmpmqaut output. Keep the previous MQSC for every changed object in the change record. For a full broker rollout plan, see the Message broker and event streaming platform rollout playbook.

Common misconfigurations

  • One SVRCONN channel and one MCAUSER for every application: every application gets the union of all permissions, and events cannot tell them apart.
  • CHLAUTH disabled to fix a 2035 error: turning off channel authentication to get an application working removes every rule at once. Fix the mapping or the OAM record instead.
  • USERMAP without credential checks: mapping an asserted user ID without CHCKCLNT(REQUIRED) or a certificate trusts whatever the client claims.
  • Weak CipherSpecs re-enabled for a partner and never removed: an AllowWeakCipherSpec line added for one legacy partner affects the whole queue manager.
  • Passwords for MQ client applications in plain configuration files: move them to a secrets manager or replace them with certificates; this is the route by which valid MQ credentials are found and reused (ATT&CK T1552.001 Credentials In Files, then T1078 Valid Accounts). If you suspect that has happened, follow the message broker compromise playbook.

Apply IBM MQ fix packs and continuous delivery updates promptly, following IBM's security bulletins for the product.

Framework mapping

IBM MQ controlNIST CSF 2.0NIST SP 800-53 Rev 5Other
CHLAUTH, CONNAUTH, per-application MCAUSERPR.AA-01IA-5 Authenticator Management, AC-3 Access EnforcementATT&CK T1078 Valid Accounts
OAM least privilegePR.AA-05AC-6 Least Privilege
TLS on channels, SSLPEERMAPPR.DS-02SC-8 Transmission Confidentiality and Integrity, SC-13 Cryptographic ProtectionATT&CK T1040 Network Sniffing
AMS integrity and privacy policiesPR.DS-02SC-8, SC-13D3FEND D3-MAN Message Authentication, D3-MENCR Message Encryption
Baseline and drift checksPR.PS-01CM-6 Configuration Settings
Authority, channel, configuration and command eventsDE.CM-03AU-2 Event Logging, SI-4 System Monitoring

For how message authentication and encryption are modelled in D3FEND, see D3-MAN Message Authentication and D3-MENCR Message Encryption.

Practitioner checklist

  • CHLAUTH is ENABLED on every queue manager and SYSTEM.DEF.SVRCONN has a no-access MCAUSER.
  • Every application has its own run-time ID, mapped by SSLPEERMAP (preferred) or by ADDRESSMAP or USERMAP with credential checks.
  • *MQADMIN is blocked on all application client channels.
  • CONNAUTH is set on every queue manager, including migrated ones, with CHCKCLNT(REQUIRED) or mutual TLS on production client channels and ADOPTCTX(YES).
  • Every channel crossing a network has SSLCIPH set to a TLS 1.3 or 1.2 CipherSpec, and no weak CipherSpec override is present.
  • OAM grants are per application, narrow, and stored as a dmpmqaut baseline in version control.
  • AMS policies protect queues carrying payment, personal or credential data, with toleration off.
  • Authority, channel, configuration and command events reach the SIEM, with alerts on blocks, TLS failures and security changes.
  • Queue manager and application certificates are in the certificate inventory with renewal owners.

Standards and references

Related NHI Mgmt Group resources: Mainframe and COBOL Application Security Guide, Message broker and event streaming platform rollout playbook, mTLS Guide, Service Account Security Guide.

PDF for subscribers

Take this guide with you

Every section, table and checklist in one PDF, with links to the sources and related guides. Subscribe to our weekly NHI & AI Identity Journal and we’ll email you a download link.

Free weekly newsletter

Subscribe to the NHI & AI Identity Journal

The latest on NHI and Agentic AI security – articles, research, breaches, news and events every week.

Bonus 33% off our NHI Course when you subscribe.

NHIMG Editorial Note
Researched and written by the NHI Mgmt Group editorial team using AI tools, under the editorial direction of Lalit Choda. Facts are drawn from the sources listed on this page and checked by our automated quality process. Last updated 14 November 2026. Spotted an error? Use our contact page.
NHI Mgmt Group, the independent authority on Non-Human Identity, IAM and Agentic AI security. nhimg.org

PDF Downloads for Newsletter Subscribers

Get the guide

Enter your details and we’ll email you a download link.

The full guide is free to read online. The PDF is a thank you for subscribing.

Fill in all fields marked * and tick the PDF box to continue.