Managed data cache of binary files for a single workflow over multiple executions.
The node should act as a gate to split the execution into a “refresh” and “loaded” path.
Similar to the loop-item-node, where the “refresh” path loops back into the cache-node.
If the “refresh” success the new-data should be saved in an well-known data directory for each workflow (exp. “/home/node/.n8n/cache/workflow-abc/node-342.bin”)
If the “refresh” failed then the old-cached-data should be used as the output
or failing, if there is no cache-data available or the cached-data is to old.
A dedicated node is required to sync-lock the “refresh” path between parallel workflow executions
and to prevent the frequent execution of the “refresh” path in a temp. failure state (backoff).
and n8n would need to clean the cache directory in one point in time.
Optional a different binary storage could be used (exp. S3),
but a local storage would already be fine.
Not to request data/config files from an external source on each execution of a workflow.
This will reduce the dependency when external source are not reachable
or the data query is costly.
Yes, but my nodejs file IO and n8n knowledge is low