My side projects were throwing maybe 200 errors a month. Sentry's team plan wanted $26/month for the privilege of telling me that. For a fleet that earns $0, that's a rounding error with a subscription attached.
So I moved everything to GlitchTip — the open-source, Sentry-API-compatible error tracker — self-hosted on Oracle's Always Free ARM VM. Six months later: $0/month, ~1,100 events captured, and migrating took one afternoon because the SDKs are drop-in.
The part that made this trivial
GlitchTip speaks the Sentry protocol. Migration is literally changing a DSN:
# before
sentry_sdk.init("https://xxx@o123.ingest.sentry.io/456")
# after — same SDK, same code, different DSN
sentry_sdk.init("https://xxx@errors.mydomain.dev/1")
No code changes. No new SDK. Breadcrumbs, stack traces, release tagging, user context — everything kept working.
Real numbers after 6 months
| Metric | Sentry (team plan) | GlitchTip self-hosted |
|---|---|---|
| Cost | $26/mo | $0 (Oracle Always Free) |
| Events/month | capped by quota | unlimited (disk-bound) |
| RAM footprint | — | ~350 MB (Docker) |
| Setup time | 0 (SaaS) | ~3 hours one-time |
| Alert channels | Slack/email | Slack/email/Telegram/webhook |
| Data residency | their cloud | my VM, my backups |
What you give up (honestly)
- Performance monitoring / APM — GlitchTip has basic perf support; Sentry's is deeper. At my scale I never opened those dashboards.
-
Zero maintenance — I run
docker compose pull && up -dmonthly. That's it, but it's not zero. - Managed uptime — if my VM dies, errors queue at the SDK level and drop after retries. Hasn't happened in 6 months.
If you're processing millions of events or need distributed tracing, pay Sentry — it's genuinely good. If you're a solo dev paying $300+/year to watch a Flask app 500 people use, you're buying enterprise insurance for a lemonade stand.
I drafted the docker-compose and the nginx reverse proxy config with MonkeyCode (free, runs against my local model — the DSNs and secrets never left my machine): https://ly.cyberserval.tech/iIETXiF
Are you self-hosting error tracking, or is Sentry's convenience worth the bill to you? Where's your break-even point — events, hosts, or just laziness?
Top comments (0)