Workflow stuck at Parallel Gateway - Intermittently

[shreyashsamani]

Hi,

We are intermittently facing a problem related to Parallel Gateway. I have attached three images, one which shows workflow model configuration details and other two which shows state of workflow i.e. stuck and successful.

Here what happen is, after successful start of workflow, execution reaches to parallel gateway which moves to 3 user tasks i.e. User Task 4 - User Task 5  - User Task 6. Now user completes User Task 5 so that particular execution moves to next node which is one input for Parallel Gateway. For User Task 4, boundary timer gets hit so it's execution also reaches another input of Parallel Gateway. Now the problem is that sometimes the execution is not moving to the next node which is Inclusive gateway.

This issue is not happening always. It has an intermittent behavior i.e. 2 times out of 10 cases. Can anybody tell me what's the problem here.

Version: Activiti - 5.15.1

Multiple(3) activiti engine with job scheduler enable in all instances.

[cjose]

Hi,

Intermittent issues are always hard to debug! 5.15.1 is quite an old version! Have you tried upgrading to one of the most recent versions on 5.x (5.22.0) or even 6.0?
If you are getting the same error even after an upgrade, you may want to dig more into the ACT_EVT_LOG & ACT_HI_ACTINST tables using the stuck process instance ids. Such an approach may help you to find out some common patterns that will help reproduce the issue. If you are successful in reproducing the issue, please raise an issue on the Activiti project with a unit test which will help the team/contributors fix the issue!

Thank you.

[shreyashsamani]

Thanks for reply Ciju Joseph‌. After analyzing logs and database entries of successful workflow and stuck workflow I found below details:

In case of successful workflow:

DEBUG [org.activiti.engine.impl.persistence.entity.ExecutionEntity] inactive concurrent executions in 'Activity(sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF)': [ConcurrentExecution[193446], ConcurrentExecution[193447]]
DEBUG [org.activiti.engine.impl.persistence.entity.ExecutionEntity] other concurrent executions: [ConcurrentExecution[193445]]
DEBUG [org.activiti.engine.impl.history.DefaultHistoryManager] Current history level: AUDIT, level required: ACTIVITY
DEBUG [org.activiti.engine.impl.bpmn.behavior.ParallelGatewayActivityBehavior] parallel gateway 'sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF' activates: 2 of 2 joined

ACT_HI_ACTINST:

ID_	PROC_INST_ID_	EXECUTION_ID_	ACT_ID_	ACT_NAME_	ACT_TYPE_
192466	192464	192464	sid-9CC591A5-8190-45B6-8545-C212DD156377	Start	startEvent
193169	192464	192464	sid-3DB1DC99-635E-4D68-8C44-BED1F6901F95	System Task 3	serviceTask
193444	192464	192464	sid-3CBE9DCB-C2F3-48F0-B04E-67F6F4BD5FE0	-	parallelGateway
193450	192464	193445	sid-CD24A1B8-EDEC-445E-8459-1994E3B00E72	User Task 6	userTask
193471	192464	193469	sid-73E2A983-15B1-4FAB-81FB-F654E5EC8609	User Task 4	userTask
193530	192464	193528	sid-1C14D193-DA7A-4F5B-8242-60F4AEB5F2BD	User Task 5	userTask
194230	192464	193447	sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF	PG2	parallelGateway
200430	192464	193469	sid-8C6396F9-8160-4FDD-997E-B72103E3B3A2	Timeout	boundaryTimer
200434	192464	193446	sid-2C4DB32B-1C81-41E5-867D-17C82F149F57	-	exclusiveGateway
200436	192464	193446	sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF	-	parallelGateway
200444	192464	193446	sid-5276E5A7-5E11-46DF-9201-483ABAF4AC69	-	inclusiveGateway
200447	192464	193446	sid-ED9A8E65-876C-447A-88DD-2D96BEC1FC6C	User Task 7	userTask

ACT_RU_EXECUTION:

ID_	REV_	PARENT_ID_	ACT_ID_	IS_ACTIVE_	IS_CONCURRENT_	IS_SCOPE_	IS_EVENT_SCOPE_	SUSPENSION_STATE_	CACHED_ENT_STATE_
193445	1	192464	sid-CD24A1B8-EDEC-445E-8459-1994E3B00E72	1	1	0	0	1	7
193446	2	192464	sid-ED9A8E65-876C-447A-88DD-2D96BEC1FC6C	1	1	0	0	1	7

In case of stuck workflow:

DEBUG [org.activiti.engine.impl.persistence.entity.ExecutionEntity] inactive concurrent executions in 'Activity(sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF)': [ConcurrentExecution[193423]]
DEBUG [org.activiti.engine.impl.persistence.entity.ExecutionEntity] other concurrent executions: [ConcurrentExecution[193422], ConcurrentExecution[193424]]
DEBUG [org.activiti.engine.impl.history.DefaultHistoryManager] Current history level: AUDIT, level required: ACTIVITY
DEBUG [org.activiti.engine.impl.bpmn.behavior.ParallelGatewayActivityBehavior] parallel gateway 'sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF' does not activate: 1 of 2 joined

ACT_HI_ACTINST:

ID_	PROC_INST_ID_	EXECUTION_ID_	ACT_ID_	ACT_NAME_	ACT_TYPE_
192428	192426	192426	sid-9CC591A5-8190-45B6-8545-C212DD156377	Start	startEvent
193114	192426	192426	sid-3DB1DC99-635E-4D68-8C44-BED1F6901F95	System Task 3	serviceTask
193420	192426	192426	sid-3CBE9DCB-C2F3-48F0-B04E-67F6F4BD5FE0	-	parallelGateway
193428	192426	193422	sid-CD24A1B8-EDEC-445E-8459-1994E3B00E72	User Task 4	userTask
193467	192426	193465	sid-73E2A983-15B1-4FAB-81FB-F654E5EC8609	For Action	userTask
193522	192426	193520	sid-1C14D193-DA7A-4F5B-8242-60F4AEB5F2BD	User Task 5	userTask
194232	192426	193424	sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF	-	parallelGateway
200401	192426	193465	sid-8C6396F9-8160-4FDD-997E-B72103E3B3A2	Timeout	boundaryTimer
200412	192426	193423	sid-2C4DB32B-1C81-41E5-867D-17C82F149F57	-	exclusiveGateway
200415	192426	193423	sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF	-	parallelGateway

ACT_RU_EXECUTION:

ID_	REV_	PARENT_ID_	ACT_ID_	IS_ACTIVE_	IS_CONCURRENT_	IS_SCOPE_	IS_EVENT_SCOPE_	SUSPENSION_STATE_	CACHED_ENT_STATE_
193422	1	192426	sid-CD24A1B8-EDEC-445E-8459-1994E3B00E72	1	1	0	0	1	7
193423	2	192426	sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF	0	1	0	0	1	7
193424	2	192426	sid-3834FD88-F389-4D28-92CB-2B4948FBBBCF	0	1	0	0	1	7

From logs of ExecutionEntity class it seems like in case of stuck workflow, one execution from User Task 5 to Parallel Gateway is still getting considered as incomplete/active though user has completed the action. Because of this condition (nbrOfExecutionsJoined==nbrOfExecutionsToJoin) of ParallelGatewayActivityBehavior class becomes false.

Please guide.

Just for reminder:

Version: Activiti - 5.15.1

Multiple(3) activiti engine with job scheduler enable in all instances.