Statsig telemetry floods with thousands of short-lived DNS queries in Termux/Android proot environments (related #8243)

Issue Description

Observed via PCAP Droid: thousands of rapid UDP DNS queries to statsig.anthropic.com, each using a new ephemeral port, even when Claude Code is idle.

Root cause: Statsig client performs aggressive retries without exponential backoff/jitter on DNS resolution failure.

Termux/proot-distro on Android routinely experiences partial/intermittent DNS (due to netd hooks, VPN, private DNS), triggering sustained bursts of failed queries.

---

Consequences on Constrained Devices

• Ephemeral port exhaustion risk (thousands of short-lived connections)
• Battery/network drain from continuous failed DNS lookups
• Connection tracker overflow in network monitoring tools
• Performance degradation on devices with limited RAM (4GB or less)

---

Related Issue

This is related to #8243 (tens of thousands of failed HTTPS requests when blocked, with no backoff).

The same pattern occurs at the DNS layer in environments with intermittent DNS resolution.

---

Requests

1. Add exponential backoff + jitter to Statsig retry logic
2. Auto-detect constrained/mobile environments and throttle telemetry accordingly
3. Make DISABLE_TELEMETRY=1 the effective default in low-resource setups (or document prominently)

---

Current Workaround

Manual opt-out via DISABLE_TELEMETRY=1 works, but default behavior punishes edge deployments and mobile users.

Hard block via /etc/hosts:

echo "0.0.0.0 statsig.anthropic.com" >> $PREFIX/etc/hosts
echo "0.0.0.0 statsig.com" >> $PREFIX/etc/hosts

---

Environment

• OS: Android 12+ (Linux 5.10.226-android12)
• Platform: Termux proot-distro Ubuntu
• Architecture: ARM64 (aarch64)
• RAM: 4GB
• Network: Mobile with intermittent DNS (VPN, private DNS, netd filtering)

---

Additional Context

The excessive DNS queries appear even when:

• Claude Code is idle (no active operations)
• Network connectivity is stable
• Other applications work normally

The Statsig client seems to retry indefinitely without proper backoff, creating thousands of DNS requests in a short period.

This behavior is particularly problematic on:

• Mobile devices with limited resources
• Networks with DNS filtering or private DNS
• VPN environments
• Environments with connection tracking/monitoring

---

Thank you for addressing this. The tool is valuable, but the telemetry behavior makes it unusable on resource-constrained environments without manual intervention.