A Cloudflare Worker that takes the usage reports and crash reports Pinky sends (docs/TELEMETRY.md, internal/telemetry) and keeps them in a D1 database. Feedback does not come here: it goes to GitHub issues.
POST /v1/usageandPOST /v1/crash: JSON, checked by src/validate.js (shape, counter names, sizes: 64 KB for usage, 1 MB for a crash); 204 when stored, 400 when refused, 413 when too large, 429 after 30 reports a minute from one address.- A usage report with the install id
00000000000000000000000000000000, or a crash report whose id starts withtest-, is checked but not stored. - Usage reports are one row per install and period: a report sent again after a failure replaces the first. Crash reports are stored once each.
- A cron at 03:23 UTC deletes usage reports older than 400 days and crash reports older than 180.
- No IP address, request header or Workers Logs are kept.
Deployed in the RSS3 account as pinky-telemetry at https://pinky-telemetry.rss3.workers.dev, with the D1 database pinky-telemetry (APAC). The app's address is telemetry.Endpoint in internal/telemetry/telemetry.go.
It has no dependencies. Use the global wrangler (not npx), from this folder:
node --test # the validator's tests
wrangler d1 migrations apply pinky-telemetry --local
wrangler dev # http://127.0.0.1:8787, with a local database
./smoke.sh # what it takes and refuses (test reports only)
PINKY_TELEMETRY_TEST_URL=http://127.0.0.1:8787 go test -run TestAgainstTheWorker ../internal/telemetry # real reports from the Go clientDeploying:
wrangler d1 migrations apply pinky-telemetry --remote # new files in migrations/ first
wrangler deploy
./smoke.sh https://pinky-telemetry.rss3.workers.devReading what came in (--local reads the local database; WRANGLER=… picks the command, which is op plugin run -- wrangler when op is installed):
./report.sh # the last 7 days: installs, versions, systems, counters, crashes
./report.sh 30 # the last 30 days
./report.sh crashes # the latest crash reports
./report.sh crash <id> # one crash with its stack
./report.sh sql "SELECT …"