Question

What are the possible reasons for stuck tokens/ tokens not moving in a active process?

Answer 1

"Stuck token" = an instance that no longer moves although nobody is waiting on purpose. The usual causes and where to look:

Cause	Symptom	Where to see it / fix
Failed system step (exception in a service)	Instance state Failed, token on the activity	Process Admin > Process Inspector > instance > error text; fix the cause, Retry (or REST action=retry)
Event manager stopped / on-hold task	Timers do not fire, system tasks after a wait do not start; instance is Active	Process Admin > Event Manager: scheduler paused? tasks on hold? resume / replay (question 2564)
Message event waiting for a message that never came / wrong correlation	Token on an intermediate message event	Check the UCA's correlation key vs the instance's variable; send the message (sendMessage) or move the token
Parallel / inclusive join waiting for a branch that ended elsewhere	Token on the join gateway forever	Model error (branch left through a boundary event) - fix the model; move / delete the token on the instance
Task assigned to a team with no members / a deleted user	Task visible to nobody	Reassign (Process Admin, REST assign), fix the team binding
Instance suspended	State Suspended	Someone suspended it (audit) - resume
Long-running service hanging (remote call without timeout)	Activity "running" for hours, thread busy	Thread dump; add timeouts; restart the member if the thread is stuck; the step fails and can be retried
Linked process / sub-process child failed	Parent waits on the linked activity	Inspect the child instance (its own id), fix / retry / terminate it
Migration to a new snapshot left the token on an activity that no longer exists	Instance failed after migration	Move the token to the successor activity; keep the old activity in future versions until instances drained
Database issues (locks, deadlocks, full tablespace)	Many instances failing at once with SQL errors	SystemOut / database logs; the instances retry after the fix

# triage from the REST API: where is the token, what is the state
GET /rest/bpm/wle/v1/process/2072.123?parts=header,executionTree,diagram      -> state, tokens (name, flowObjectId, createdTaskIDs), steps
PUT /rest/bpm/wle/v1/process/errors?instanceIds=123                            -> runtime error messages of failed instances
GET /ops/std/bpm/event_manager_tasks?process_id=123                           -> pending timers / async steps of the instance (BAW 20+)

A dashboard that lists Failed instances, on-hold event manager tasks and tokens older than N days catches most of these before users notice.

References

• Process REST API - executionTree, errors, retry, moveToken
• Operations REST API - event manager tasks
• BAW documentation - Process Inspector and Event Manager