refactor(connd): use new datadog client

[vmenge]

changes

• connd to use shared datadog client

[github-actions]

Found two issues:

1. orb-connd/src/main.rs:116 switches connd to orb_dogd::DogstatsdClient, but that client only waits for /run/datadog/dsd.socket and no longer preserves the old UDP fallback. The deleted local client explicitly fell back to dogstatsd::Options::default() when the socket was absent, so on systems where DogStatsD is reachable over UDP but the UDS is missing, connd will stop emitting metrics.

2. orb-connd/src/reporters/datadog.rs:69, :76, :94, :100 and orb-connd/src/reporters/data_usage.rs:47, :55 now discard MetricEmitter errors. orb-dogd returns errors when its bounded queue is full or the worker dies; the old code propagated these with ?, so the reporter logged/restarted. Please at least log these failures, otherwise metric loss from backpressure/worker failure becomes invisible.

I couldn’t run cargo check in this sandbox because rustup failed to create temp files on the read-only filesystem.

[vmenge]

Found two issues:

1. orb-connd/src/main.rs:116 switches connd to orb_dogd::DogstatsdClient, but that client only waits for /run/datadog/dsd.socket and no longer preserves the old UDP fallback. The deleted local client explicitly fell back to dogstatsd::Options::default() when the socket was absent, so on systems where DogStatsD is reachable over UDP but the UDS is missing, connd will stop emitting metrics.
2. orb-connd/src/reporters/datadog.rs:69, :76, :94, :100 and orb-connd/src/reporters/data_usage.rs:47, :55 now discard MetricEmitter errors. orb-dogd returns errors when its bounded queue is full or the worker dies; the old code propagated these with ?, so the reporter logged/restarted. Please at least log these failures, otherwise metric loss from backpressure/worker failure becomes invisible.

I couldn’t run cargo check in this sandbox because rustup failed to create temp files on the read-only filesystem.

two issues according to whom

plz evaporate