feat(collector): per-process benign-noise filter for known transient sources

e29519581fc7c8a12e149487ec448c5d6a97e969 · 2026-05-11 20:44:59 -0700 · SteveStudio2

Files touched

M index.js

Diff

commit e29519581fc7c8a12e149487ec448c5d6a97e969
Author: SteveStudio2 <stevestudio2@SteveStacStudio.lan>
Date:   Mon May 11 20:44:59 2026 -0700

    feat(collector): per-process benign-noise filter for known transient sources
---
 index.js | 29 +++++++++++++++++++++++++++++
 1 file changed, 29 insertions(+)

diff --git a/index.js b/index.js
index 4294b1a..6330e82 100644
--- a/index.js
+++ b/index.js
@@ -47,6 +47,29 @@ const PM2_LOG_DIR = path.join(os.homedir(), '.pm2', 'logs');
 
 function log(...a) { console.log(new Date().toISOString(), ...a); }
 
+// Per-process patterns that match expected, non-actionable log noise.
+// If EVERY non-empty line in a new log chunk matches the regex, we skip posting.
+const BENIGN_NOISE_PATTERNS = {
+  // cloudflared rotates edge connections every ~16s; the reconnect cycle logs
+  // ERR but it's normal heartbeat behavior, not a failure.
+  'big-red-tunnel': /Tunnel connection curve preferences|failed to run the datagram handler.*context canceled|failed to serve tunnel connection.*control stream encountered|Serve tunnel error.*control stream encountered|Retrying connection in up to|Failed to dial a quic connection.*handshake did not complete in time|^\{.*"level":"(info|error)".*\}$/,
+  // microsite-hawk emits cycle-complete on success — should be stdout, but
+  // belt-and-suspenders in case the source fix regresses.
+  'microsite-hawk': /^\[hawk\] Cycle complete .* failures: 0/,
+  // abrams-report scrapers log vendor HTTP 4xx/timeouts as warnings; these are
+  // vendor-side issues, not actionable code bugs.
+  'abrams-report-scrape': /^\[(html|rss)\] [a-z0-9-]+ failed: (HTTP 4\d\d|Request timed out|network timeout)/,
+  'abrams-report-wallpaper': /^\[(html|rss)\] [a-z0-9-]+ failed: (HTTP 4\d\d|Request timed out|network timeout)/,
+};
+
+function isBenignNoise(processName, chunk) {
+  const pat = BENIGN_NOISE_PATTERNS[processName];
+  if (!pat) return false;
+  const lines = chunk.split('\n').map(l => l.trim()).filter(Boolean);
+  if (lines.length === 0) return true;
+  return lines.every(l => pat.test(l));
+}
+
 function loadState() {
   try {
     if (fs.existsSync(STATE_FILE)) return JSON.parse(fs.readFileSync(STATE_FILE, 'utf-8'));
@@ -260,6 +283,12 @@ async function scanLaunchdErrLogs(state) {
     if (!meaningful) { state.logOffsets[errPath] = cur; continue; }
 
     const processName = label.replace(/^com\.steve\./, '');
+    // Per-process known-benign-noise filter. Each entry: if EVERY non-empty line
+    // in the chunk matches the regex, advance the offset without posting.
+    if (isBenignNoise(processName, chunk)) {
+      state.logOffsets[errPath] = cur;
+      continue;
+    }
     const posted = await postTask({
       source: 'launchd',
       processName,

← 064993f tick 9: cross-platform — also runs on Kamatera, host-tagged · back to Cncp Failure Collector · feat(collector): also filter nginx TLS handshake bot noise b3aa723 →