The service result cache is per JVM and keyed by service + inputs (question 2897). Ways to reset it without restarting the server:
- Install a new snapshot of the app (even with no change): the caches of the old snapshot's services are dropped, the new snapshot starts cold. In development, taking a new snapshot / redeploying the tip on the Process Center's playback server does the same.
- Process Admin > Admin Tools > Manage Caches (8.5.7 / 8.6; under Server Admin on BAW) lets you clear the engine caches, service caches included, per server; on a cluster do it on every member.
- Let the lifetime expire: the cache entry lives for the number of seconds set on the service (Cache results for); set a short lifetime on services whose data changes during the day and a long one for reference data.
- Change the inputs: because the key includes the inputs, adding a harmless "cache buster" input (a version number from an EPV) makes a new cache entry when the EPV changes - a trick for reference data reloaded after an admin action.
# BAW 20+ / CP4BA: engine caches are also reachable through the Operations REST API
# check /ops/docs of your level for a cache resource; the fallback that always works: re-activate the snapshot
POST /ops/std/bpm/containers/{acronym}/versions/{snapshot}/activateWhat does not help: clearing the browser cache (the service cache is server-side) and "Refresh" in the designer. If you find yourself clearing caches often, the service should not be cached - cache only truly static lookups.
References