[Divergence] New powersync_sync_connections_total Prometheus metric (powersync-service #786)

[benitav]

What's changing: The Service adds a new metric, powersync_sync_connections_total, counting sync stream connections once at close. It's labeled by outcome, close reason, PowerSync error code, and transport, so operators can spot error spikes and measure sync reliability without parsing logs.

High priority — existing docs are now inaccurate or incomplete

• Usage Reporting (https://docs.powersync.com/maintenance-ops/self-hosting/usage-reporting), "What is Collected" table: the table enumerates every dimension from the packages/types/src/metrics.ts file it cites as its source, including concurrent_connections. The new powersync_sync_connections_total metric is now in that file but missing from the table.

For reviewer consideration — worth documenting

• Monitoring (https://docs.powersync.com/maintenance-ops/self-hosting/monitoring): describes the self-hosted Prometheus endpoint generally but doesn't name individual metrics; worth checking whether it should point to this new metric or its close-reason/error-code/transport labels for diagnosing connection failures.

Source PR: powersync-ja/powersync-service#786 — open, labeled "Product Visibility".

Filed automatically by Claude Code (claude-sonnet-5). A human must verify the scope, prepare the docs PR, and publish it only once this update has been released.