Open sczyh30 opened 3 years ago
From my perspective, maybe a global config for monitoring mode is enough?
Maybe two levels are needed: global level and rule level. A new kind of metric (like "should block") is required for monitoring mode. More discussions are needed for details (e.g. event propagation, metric logs).
I think monitoring mode is necessary. For production safety, we usually want to configure a monitoring mode rule first
Issue Description
Type: feature request
Describe what feature you want
RFC: Support monitoring (dry-run) mode. In monitoring mode, Sentinel won't actually block the requests, but only record the relevant metrics (for observation). This could be useful when users need to verify whether the rules are appropriate.