🟠 LOW SEV · Redis Command Latency High
🔥 Redis command p99 is over 1ms — this latency is on the user's connection path
<|📈 graph> · <|📊 service dashboard> ·
🚨 alarm definition ·
🔕 silence
_Every new connection through the tunnel does an inline proxy:block lookup, and every first-time DNS resolution does a mock-IP allocation. Both are synchronous Redis calls, so command latency here is added directly to connection setup for the child's device.
Redis is single-threaded, so this is usually a consequence rather than a cause — check 'Redis Fork Stall' and the slow-log rule first, then memory pressure.
Baseline: p99 is 2 MICROSECONDS. The 1ms threshold is ~500x that, deliberately far out so it only fires on a real change in kind, not on jitter.
Panel: /d/redis-service-health 'Is anything stalling the event loop?'. Ticket: CLE-514._
Grafana vmain-b1e6358