(OO) Support Tip: Java.lang.OutOfMemoryError: GC overhead limit exceeded and java.lang.OutOfMemory
How to resolve Java.lang.OutOfMemoryError: GC overhead limit exceeded and java.lang.OutOfMemoryError: Java heap space errors in this environment?
In OO 10.5x, quartz was upgraded. Engineering is not able to reproduce the problem in our lab. We assume that the problem was fixed by upgrading quartz because of two reasons: (1) OO does not control the quartz trigger lifecycle. This is 100% managed by the library. (2) Other customers who reported this issue and since upgraded to 10.51 have no longer reported it. Technical Support recommends upgrading this environment to OO version 10.51.
This issue was seen in a customer environment where after performing the workaround steps below the issue was no longer seen. The root cause was the blocked splitJoinTrigger which caused a lot of unfinished run data to pile up in the database and to be loaded to the Central run engine context. As a result of the blocked splitJoinTrigger issue, defect QCCR8C24579 splitJoinTrigger found in BLOCKED state was submitted. Defect QCCR8C24579 was cloned to defect QCCR8C28538 splitJoinTrigger found in BLOCKED state. A hotfix was created for OO version 10.22. Please contact Technical Support for the hot fix for version 10.22 and if you have any other questions.
1. Stop Central services.
2. Clean all active run remains from the DB:
TRUNCATE TABLE OO_BRANCH_CONTEXTS;
TRUNCATE TABLE OO_FINISHED_BRANCHES;
DELETE FROM OO_SUSPENDED_EXECUTIONS;
TRUNCATE TABLE OO_EXECUTION_QUEUES_1;
TRUNCATE TABLE OO_EXECUTION_STATES_1;
TRUNCATE TABLE OO_EXECUTION_STATES_2;
DELETE FROM OO_EXECUTION_SUMMARY WHERE STATUS NOT IN (‘COMPLETED’,'CANCELED',’ SYSTEM_FAILURE’);
3. Set manually the splitJoinTrigger to WAITING in the OO_TRIGGERS table:
update OO_TRIGGERS set TRIGGER_STATE='WAITING' where TRIGGER_NAME='splitJoinTrigger'
4. Start Central Services.
Please see the knowledge document at https://softwaresupport.hpe.com/km/KM02200416