Hello, please find below a description of an issue I am facing with my alert consumer on a very simple filter :
Distribution: Kafka stream, with lite lightcurve
WHERE: objects.nDiaSources > 1 AND (sherlock_classifications.classification != ‘VS’ OR sherlock_classifications.classification IS NULL)
Tables joined: objects, sherlock_classifications, crossmatch_tns
The filter is actively matching real, fresh DiaObjects (confirmed via the /api/query/ REST endpoint), but those matches are not being published to the filter’s Kafka topic. Only a static, unchanging batch of ~10 recycled example objects ever arrives on the topic, delivered once whenever the filter is re-saved in the web UI but never as new matches occur.
Is there a known issue where a filter’s live/incremental Kafka delivery stops working (while the filter’s match query itself keeps working, and the one-time save-triggered test push keeps working)? Is there anything filter-side (e.g. re-activating the filter, a distribution-settings toggle, or the join across objects/sherlock_classifications/crossmatch_tns) that could cause this?
Happy to provide the filter ID/account details via a non-public channel if useful for debugging.