Exception in completeTask

[satramalin]

Hi,
Can you please let me know when this exception happens
Exception Occured in completeTask(): HistoricVariableInstanceEntity[id=1907296, name=request, revision=1, type=serializable, byteArrayValueId=1907297] was updated by another transaction concurrently
Thanks!!

[trademak]

I would expect that the same variable is changed by multiple transactions at the same time. Would that be possible in your implementation?

Best regards,

[satramalin]

We don't have that scenario in our implementation. Its a simple test process which has 3 user tasks assigned to the same user.  We are checking the load by creating large number of instances in a high speed. When the load increases, we observed this exception. We also observed the below exceptions.

1. Exception Occured in completeTask():"Exception Occured in completeTask():
### Error updating database.  Cause: java.sql.SQLException: ORA-00060: deadlock detected while waiting for resource
### The error may involve org.activiti.engine.impl.persistence.entity.HistoricActivityInstanceEntity.updateHistoricActivityInstance-Inline
### The error occurred while setting parameters
### SQL: update ACT_HI_ACTINST set       EXECUTION_ID_ = ?,       ASSIGNEE_ = ?,       END_TIME_ = ?,       DURATION_ = ?     where ID_ = ?
### Cause: java.sql.SQLException: ORA-00060: deadlock detected while waiting for resource

2. Exception Occured in completeTask(): Cannot find task with id

Thanks

[jbarrez]

A deadlock is surely something else than the optimisticlocking exception.

What are you testing here exactly? How many instances? What kind of Oracle database?

[satramalin]

We are performing a load test on Oracle exadata. There are 3500-4000 instances created for a single run. Right now, there are more than 100000 instances in the database. Out of close to 3500 instances in every run, we observed the exceptions for 10-15 instances. Most of the exceptions are with concurrent updates.

Thanks

[satramalin]

Please find the attached bpmn

[hepcibha]

Hi, we are having a similar issue in production, did you find a solution for this?

[jbarrez]

You mean a deadlock? We would need more info about the process and how to reproduce.