One stream, one reader: designing a standby that takes over
A second connection with the same key closes the first, so redundancy means failover, not parallel readers. A standby design that actually works.
PNCLPULSE
Dated notes on turning Pinnacle odds API drops into alerts: thresholds, streams, relays and the mistakes that cost a signal.
A second connection with the same key closes the first, so redundancy means failover, not parallel readers. A standby design that actually works.
min_drop filters the size of the fall. Your rules can also filter where the price landed: using the nvp field to keep only the alerts that matter.
Not every alert deserves its own ping. Buffer the stream into a window, group by match, and send one message that reads like a report.
The stream handles reconnects; the reader process needs its own care. Restart policy, a silence watchdog, state on disk, and an alarm for the reader itself.
recheck=N holds each prematch drop for N seconds and sends it only if the price has not bounced back. What it filters, what it costs, how to set it.
Save the data: lines from one live session, replay them through your relay from a local file, and watch the filter, dedupe and formatting work.
One stream, many destinations. Put a routing table between your filter and your webhooks so each sport and severity lands in its own channel.
min_drop is set per connection and you get one connection per stream. Set the floor low, then apply per-sport thresholds in your own code.