SYMPTOM
In a clustered Strategy Intelligence Server environment, during report execution, History List status does not appear to be synchronized between the nodes.
When a report is executing, the node on which the job is running may show a status of 'Executing' in the History List while the other nodes may show a status of 'Waiting' in the History List.
CAUSE
This is working as designed for performance reasons. The History List is only synchronized at the beginning and the end of a job as well as if the job errors out or if it is terminated. The History List is not synchronized during the execution of the job.
The following steps illustrate this issue better:
- Consider a cluster of Strategy Intelligence Server with two nodes, A and B. Configure the servers to use only one thread to the warehouse.
- Send five reports to the History List from Strategy Desktop pointing to Node A.
- The History List on Node A indicates one job in 'Executing' status and four jobs in 'Waiting' status because there is only one thread configured.
- The History List on Node B indicates all five jobs in 'Waiting' status. The status of these jobs remains as 'Waiting' even after purging the client-side object cache and disconnecting/reconnecting to the project source. This is by design.
- When the report is done executing, the History List on Node A indicates one job in 'Ready' status, one job in 'Executing' status and three jobs in 'Waiting' status. The History List on Node B indicates one job in 'Ready' status and four jobs in 'Waiting' status.
- If the report does not finish because it errors out or is terminated, then the History List on Node A indicates this job with 'Execution error' status. The History List on Node B will also indicate this job with 'Execution error' status.
KB7777