We don't currently have a way to push actions to Bouncer
This would be helpful in dealing with URLs we later find to be a false positive or false negative
In those cases, we would want to lookup the servers where those URLs were misclassified and "tell" Bouncer to treat the old messages with those URLs correctly
This relates to #93
Because we can't push actions, we can do the distant next best thing, which is to add an action to a response for an unrelated request. Here's the flow:
Bouncer from a given server queries BE with a URL
BE tells Bouncer the URL is safe
Bouncer gets the response and does nothing
BE later re-scans the URL and finds it malicious
Bouncer later queries BE with an unrelated URL
BE tells Bouncer whether the unrelated URL is safe AND that the previous URL is unsafe
BE will send the URL and the message IDs associated with the URL
Bouncer gets the response and:
Ignores the unrelated URL because it's safe, and...
Goes through the message IDs and, if it finds the unsafe URL in those messages, deletes them as normal
NOTE: We'd need a way for a given Bouncer instance to send a message back to BE that a given re-classified URL has been correctly dealt with.
BE would then need a way to track that that message ID(s) on that server are no longer misclassified
We don't currently have a way to push actions to Bouncer
Because we can't push actions, we can do the distant next best thing, which is to add an action to a response for an unrelated request. Here's the flow:
NOTE: We'd need a way for a given Bouncer instance to send a message back to BE that a given re-classified URL has been correctly dealt with.