[ziyan]

Summary

Remote mesh nodes showed SQL p50 ~5x the raw RTT: the default pool (2 idle,
unbounded open) makes bursty paths repeatedly pay the peer tunnel setup cost
(tcp + ssh + tls + postgres auth ≈ 1.5-2.5s), folded into whichever statement
triggered the dial.

• keep 8 idle connections warm, for up to 5 minutes
• cap open connections at 16 per node (4 nodes ≤ 64, postgres now at
  max_connections=200)

Changed

• Database connection pool keeps idle connections warm and bounds open
  connections, avoiding repeated peer tunnel setup costs on remote nodes

Test plan

• make test (postgres + -race), make lint, make lint-mulint all clean
• config-only change; the effect will be measured on the live mesh (expect
  remote SQL p50 to drop from ~850-980ms toward the ~1x RTT floor)

[codecov]

Codecov Report

❌ Patch coverage is 66.66667% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 62.28%. Comparing base (4a5f2ba) to head (e565ac0).
⚠️ Report is 3 commits behind head on master.

Files with missing lines	Patch %	Lines
db/db.go	66.66%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #16      +/-   ##
==========================================
- Coverage   62.49%   62.28%   -0.21%     
==========================================
  Files          28       28              
  Lines        2293     2299       +6     
==========================================
- Hits         1433     1432       -1     
- Misses        688      692       +4     
- Partials      172      175       +3     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.