The MREF Admin Console, page by page

Getting inOpening the console

Sign in to MREF as an administrator, then open https://<your MREF host>/html/en/default/admin/index.jsp in the same browser. The left menu lists the pages you are allowed to see; the right side shows the selected page.

Access is per page, per user: see Admin Users below. A user with "Read Only" on a page can look but not press its buttons.

The big pictureHow MREF runs on OpenShift

On Maximo Application Suite, MREF is not one server. The Facilities operator builds it from a custom resource (the FacilitiesWorkspace) into several pods in its own namespace. On this system, mas-instdb2-facilities:

So when the Admin Console says an agent is "Running On instdb2-wsp-multiagents-3000", it means the multiagents pod. Its log (/home/wiotp/log/server.log inside that pod, or the pod's Logs tab in OpenShift) is where that agent's errors go, not the Error Logs page of the UI server.

Page 1Admin Summary and the red warnings

The red System Status text

The two messages above the menu on this system:

• "Maximo Application Suite session timeout setting could not be retrieved." MREF asks MAS for the session timeout so both log users out at the same time. The server log shows the call is refused (HTTP 401). Users can still work; MREF falls back to its own timeout (System Info → User Session Timeout, 30 minutes here). Worth fixing so sessions behave consistently.
• "Invalid AES Encryption Keystore Password detected. Refer to the vaultSecret setting…" MREF encrypts some stored values with an AES key kept in a keystore. Its password comes from the vault secret named in the FacilitiesWorkspace (vaultSecret). The warning means that password does not open the keystore. Fix it in OpenShift, by correcting the secret, never by editing anything in the console. Until then, features that rely on encrypted values may fail.

The heart of itAgents, the background workers

An agent is a background process inside a Liberty server that wakes up regularly, looks for queued work in the database, and does it. Almost everything that is not a user clicking a button is done by an agent: asynchronous workflows, scheduled events, imports, publishing, cleanup.

Reading the statuses

• Running with a server name: normal.
• Not Running: not started. Fine for agents you do not use (SNMP here).
• Start Failed: something tried to start it and it could not. Here the Reserve SMTP agent is switched on for the multiagents pod but Reserve mail is not configured (its domain is still the sample reserve.tririga.qa.tririga.com). Harmless if you do not use Reserve; otherwise fix the configuration.

Where an agent runs, and why

On OpenShift the multiagents pod is started with agent switches such as WF_AGENT=enabled, SCHEDULER_AGENT=enabled, MAINTENANCE_AGENT=enabled and RESERVE_SMTP_AGENT=enabled; the operator sets them from the FacilitiesWorkspace (agentsSpecs, empty here, means "use the defaults"). Each Liberty server has an instance ID (0 for the appserver, 3000 for multiagents), which is the number you see after the pod name.

Tuning the agentsPages that configure agents

Thread Settings

How many things each agent may do at once. More WFAgent threads means more workflows in parallel, but also more database connections and memory. These values come from FACILITIES.properties (WFAgentMaxThreads=80 and so on); on MAS, change them there, not here.

Workflow Agent Info

• User and Group List: you can dedicate a workflow agent to certain users (for example, a heavy integration user). Not used here.
• Run Workflows Triggered By Scheduled Events As: the user that scheduled workflows run as (system).
• Workflow Instance Recording: Errors Only: instances are kept only for failed workflows. Switch to "All" briefly when you debug a new workflow, then back: recording everything fills the database.
• Workflow Max Threads 80, Max Active Threads Per User 70: the per-user limit stops one user (or one integration) from taking every thread.

Scheduler Info

A recurring schedule with no end date would create endless events, so the scheduler only creates the next 100 daily or weekly, 50 monthly or 10 yearly occurrences, and tops them up later.

Maintenance Agent

A small scheduler for database housekeeping: here Db2 table and index reorganisation (TRI_DO_REORG_TABLE_AND_INDEX) every Sunday. The cleanup itself (old workflow instances, stale records) runs daily at CLEAN_HOUR from FACILITIES.properties (0 here, midnight UTC). Agree this with your DBA: on MAS the database is often managed separately.

Reserve SMTP Agent

Watching the workWorkflow Events, Workflows Executing, Performance Monitor

These are the pages to open when "a workflow did not run" or "the system is slow".

• Workflow Events in the Queue: pending events per user. Empty is good. A number that only grows means the WFAgent is stopped, overloaded or failing.
• Oldest Event in Queue: how far behind the agent is.
• Future Actions: actions waiting for a date (WFFutureAgent).

A workflow that never finishes (a loop, a lock) shows here with its current step and run time. Stop it here rather than restarting the server.

The quickest health check: CPU, heap, open database connections, workflow queue count and throughput, workflows started, completed and failed (since the server started), extended formula queue and logged-in users. "Record to Log" writes the numbers to the log, handy before and after a load test.

DataDatabase Info, Database Query Tool, Data Connect, Caches

Database Info

The top checks that the database and server clocks agree, and that the database is UTF-8 (CODEUNITS32 on Db2). The Database Admin Tasks run the cleanup jobs now instead of at night: clean up workflow instances, stale records, scheduled events, re-analyse statistics, check the organisation and geography hierarchies, deep cleanup, check and restore associations. They change data and can run for a long time: use them on purpose, outside business hours, with a recent backup.

Database Query Tool

Runs SELECT statements only. The list on the left is a library of ready-made audit queries (agent settings, record counts, workflow queue counts, logins). It is the fastest way to answer "did the workflow run?" (WF_EVENT_HISTORY) or "what is in this record?" (the T_<BO> table).

Data Connect

When an external system writes rows into staging tables, DataConnect turns them into records in batches ("jobs"). This page shows each job's state and lets you retry or fail it. Empty here: no DataConnect integration is running.

Caches

MREF keeps metadata in memory for speed. When a change does not show up, flush the matching cache rather than restarting: Workflows For Agent after publishing a workflow the agent does not pick up, Security Scope after organisation or group changes, Query Cache after query changes, Object MetaData Cache after Data Modeler work. All Caches (Global) flushes everything on every server; it is safe but makes the next minutes slower for everyone.

DiagnosticsError Logs, Platform Logging, Metadata Analysis, Java Info, Must Gather

Error Logs

Click a log to read it, "Roll" to start a fresh file. server.log is the main one. Remember it is the log of the UI server; agent errors are in the multiagents pod's log.

Platform Logging

Tick a category to add DEBUG output for it to the log: for example Workflow Agent → Event check/pickup when workflows do not start. Untick it when you are done: debug logging is large and slows the server.

Metadata Analysis

Renders every form, runs every query, loads every workflow, generates publish DDL. Useful after an upgrade or a big migration to find broken metadata. Each one is heavy: run it on a test system or at night.

Java Info

Note the warning on Force Garbage Collection: it can pause the server. Leave it alone.

Must Gather Tool

Use it when IBM Support asks for it. On MAS, also collect the OpenShift side (the MAS must-gather) so support sees the pods too.

SystemSystem Info, Build Number, Licenses, Admin Users, Users Logged In

System Info

• Property Files: view the TRIRIGA properties and Liberty server.xml of this pod. On MAS, edit them through the FacilitiesWorkspace secrets, not here.
• Lock System: blocks new logins, for example during an Object Migration. Sign in to the application first, or you lock yourself out.
• Username and Password Encryption: encrypts a value for use in a properties file.
• User Session Timeout: 30 minutes here (the fallback while MAS's value cannot be read).
• Feature Flags: switches for newer behaviour, such as the Carbon-style record pages.

Build Number

The first page IBM Support asks about. Code build, database build and language pack should all match.

Licenses

On MAS, who gets which licence (and AppPoints) is decided in Suite Administration; this page only shows what each licence type unlocks.

Admin Users

Two pages, Platform Properties and AES Encryption, are not ticked even for system: they are hidden from the menu until someone grants them. Keep it that way unless you need them.

Users Logged In

Lists active sessions (user, IP address, login time, last action) with a "Log Out User" action. Use it before locking the system or deploying a migration package. (No screenshot: it shows session identifiers.)

Before you clickSafe, careful and dangerous buttons

GlossaryWords you will hear

Agent

A background worker inside a Liberty server that polls a queue and does the work. Maximo: a cron task instance.

Instance ID

Number of a server in the cluster: 0 for the UI pod, 3000 for multiagents here.

Multiagents pod

The MAS pod that runs the agents and serves no users.

FacilitiesWorkspace

The OpenShift custom resource the operator reads to build MREF. Where settings really live.

FACILITIES.properties

The TRIRIGA properties (threads, cleanup hour, workflow recording), delivered as a secret.

Vault secret

The secret holding the password for MREF's AES encryption keystore.

Workflow event

A queued request to run an asynchronous workflow (table WF_EVENT).

Future action

A workflow action waiting for a date (table WF_EVENT_FUTURE).

Cache flush

Throwing away in-memory metadata so it is reloaded from the database.

Must Gather

A diagnostics package for IBM Support.

Screens: IBM Maximo Real Estate and Facilities 9.2.2 (platform 9.2.6) on Maximo Application Suite and Red Hat OpenShift, test system, viewed read-only. Agent descriptions summarise IBM's documentation and what this system shows; check IBM's documentation for your version before changing anything.