Problem Definition

In Jira Service Desk Server 3.4.1 an SLA Audit Log was introduced for troubleshooting SLAs. This audit log can grow to huge sizes in the database, so a cleanup job was introduced(https://jira.atlassian.com/browse/JSDSERVER-5266). This cleanup job helps to keep the size of the database tables smaller, but can still allow the audit log to grow to large sizes and the job itself can become very resource intensive and long running. This leads to a performance impact due to the SQL query used and a stability impact because the query can become long running(over 30 hours has been seen) and can block other database calls.

Suggested Solution

Provide a dark feature, or other configurable parameter, to completely disable the SLA Audit Log.

Why this is important

The tradeoff between the benefits of the SLA Audit Log and the impact it can create is something that may vary from instance to instance and should be configurable to best fit the use case for that instance.
This is important because admins should be given the choice of whether to use the audit log feature or not.

Workaround

 
The closest thing to a workaround is to set a configuration setting in the database to reduce the number of days before an audit event is cleaned up. The default is 30 days and this can allow the tables to get too large, slowing down the cleanup job. If we reduce this to 1 day the tables will be kept much smaller and because they are smaller the cleanup job is able to complete much more quickly.

The database property is:

sd.sla.audit.log.cleanup.days

Run the SQL query to determine the ID of the property in the database:

SELECT * FROM propertyentry JOIN propertynumber ON propertyentry.ID=propertynumber.ID WHERE PROPERTY_KEY = "sd.sla.audit.log.cleanup.days";

Then to modify this parameter from 30 to 1:

UPDATE propertynumber SET propertyvalue=1 WHERE ID = <ID from above>;

Related tickets:

• https://jira.atlassian.com/browse/JSDSERVER-5266
• https://jira.atlassian.com/browse/JSDSERVER-5719