SYMPTOM:
In Strategy 9.x and 10.x clustered environment, multi-recipient report subscription jobs execute against the warehouse on each node instead of the job on the second node waiting for a cache to be created by the first node.
For example, if jobs 1, 2, 3, 4 all execute on node A, then job 1 will run against the warehouse while jobs 2, 3, 4 will wait for the cache to be generated. However, if jobs 1, 2 execute on node A while 3, 4 on node B, then jobs 1 and 3 will run against warehouse and only 2 and 4 will wait for cache instead of having jobs 3 and 4 also wait for cache.
CAUSE:
This is working as intended. Waiting for cache is not meant to work across nodes in a cluster. Waiting for cache does not work across the cluster, only within a single node.
WORKAROUND:
Ensure that a full report cache already exists before running the subscription by manually running the report, or scheduling one recipient subscription ahead of the others. This way, only one job will run against the warehouse and subsequent jobs will use that cache.
Related reading:
KB40743 - In Strategy Intelligence Server 9.x and 10.x, multi-recipient subscriptions created for reports execute multiple report jobs instead of waiting for report cache.
