In our production environment we have multiple web player nodes each with two instances.
We have a number of critical reports that we run as a scheduled update where we select 2 in the 'number of instances' dropdown.
The issue with default routing is that it can sometimes cache the report on the same node twice on each of the two instances. If that node goes offline the report is down/unavailable. This happens to us now sometimes in production.
There are options to mitigate such as resource pools but they need constant monitoring by admins as reports can grow in size over time and we'd prefer to keep default routing.
Idea for an enhancement request would be to add a checkbox to the right of the number of available instances label on the scheduling UI that would be named “Cache on different web player nodes” which when checked would ensure cache is built on different nodes regardless of how many instances are defined to the node. Seems like a simple thing to implement yes?
Dave, thanks for the comment and yes that is indeed the workaround we are implementing but it is a workaround. The enhancement request would allow customers to avoid this trouble and stay with default routing.
Can't you achieve this if you have multiple services on each node and then create a resource pool that uses one service from each server node would be used if you select to cache to a shared resource?