There are no metrics for feeds. pkg/feeds has no metrics.go, feedGetHandler in pkg/api/feed.go is uninstrumented, and the /feeds/{owner}/{topic} route is mounted without the metric middleware that /bzz and /bytes use.
Effect: on a production node there is no way to measure real feed lookup volume, failure rate or latency — feed traffic can't be told apart from ordinary chunk retrieval.
A feed lookup histogram would cover it, ideally with labels for lookup type (sequence/epoch) and result (found / not_found / error).
Synthetic beekeeper checks give some coverage today, but they only exercise test traffic and can't tell us anything about real users.
There are no metrics for feeds.
pkg/feedshas nometrics.go,feedGetHandlerinpkg/api/feed.gois uninstrumented, and the/feeds/{owner}/{topic}route is mounted without the metric middleware that/bzzand/bytesuse.Effect: on a production node there is no way to measure real feed lookup volume, failure rate or latency — feed traffic can't be told apart from ordinary chunk retrieval.
A feed lookup histogram would cover it, ideally with labels for lookup type (sequence/epoch) and result (found / not_found / error).
Synthetic beekeeper checks give some coverage today, but they only exercise test traffic and can't tell us anything about real users.