So we cache tokens for a given time and as long as the session is active token is checked in cache and all is good. We also have an API that invalidates session/token. Now i am able to clear cache entry on the server that the request lands on. If the next request with the same token lands on the same server the request fails as expected since cache is empty but if it lands on any other server within the cluster it doesnt fail since the cache is NOT cleared.
Is there a way around it?
For the scenario that you have provided I would suggest that you look to use the Tactical Remote Cache assertion as the localized cache will not be able to update all the nodes in the cluster. This assertion is not provided out of the box. However, if you raise a support case you can request access to it. There is no separate license needed to use it.
Thanks Stephen. Will try that out.