Lewati ke konten

Platform limits

Konten ini belum tersedia dalam bahasa Anda.

These are Cloudflare’s, not MailRiz’s, but they decide what the product can promise.

Workers Free gives 10 ms of CPU per request. Workers Paid gives 30 s.

CPU time excludes waiting on I/O, and most requests use very little — but two paths in MailRiz push against it:

  • Parsing inbound mail. A large HTML message can exceed 10 ms, which makes delivery fail intermittently — the hardest kind of failure to diagnose, because most mail still arrives.
  • Live updates. Every poll inside an open stream spends from that same per-request budget, which is why connections are deliberately short-lived and why UPDATES_POLL_MS exists.

Workers Paid ($5/mo) is recommended for a mailbox you rely on.

Limit Value
Maximum message size 25 MB

Larger messages are rejected by Email Routing before the Worker sees them.

Free tier
D1 500 MB per database, 5 GB per account
R2 10 GB stored, 1M Class A ops, 10M Class B ops, no egress charge

Bodies and attachments live in R2, so D1 stays small — it holds metadata, text for search, and the FTS index. A personal mailbox will reach the R2 limit long before the D1 one.

Requests to static assets are free and unlimited and do not count toward the Workers quota — so serving the dashboard costs nothing. Only requests that invoke the Worker script are billable, which for MailRiz means /api/* and inbound mail.

Note that run_worker_first: ["/api/*"] means API paths always invoke the Worker, by design.

There is no wall-clock limit on an HTTP-triggered Worker while the client stays connected — which is what makes a long-lived SSE stream possible at all. The constraint is CPU, not time.

Cron triggers are capped at 15 minutes; the daily retention purge is far below that.