test(scale): measure big lists and big archives, document the numbers

- Add make scale: a 10000-member fan-out through the queue and a 50000-message import with search latencies
- Report the figures in the README
This commit is contained in:
tgiachi 2026-11-09 20:32:09 +01:00
commit e4e08cd25e

View file

@ -220,6 +220,21 @@ lists.example.org lmtp:inet:127.0.0.1:2424
- The API and the admin UI use one bearer token and no TLS: put them behind a reverse proxy with TLS and keep the `api`
listener private.
## Scale
Measured on one machine with `make scale` (SQLite file on a local disk, relay on localhost; your numbers will differ):
| | |
|---|---|
| Queue a post for a 10 001-member list (one transaction) | 0.12 s |
| Send it through a local relay, 4 deliveries at a time | about 460 messages/s (22 s for the whole list) |
| Import a 50 000-message mbox (14 MiB) | 2.9 s, about 17 000 messages/s |
| Search a 50 000-message archive (one page of 25) | 6 to 15 ms |
| Open page 1 000 of the archive, read a thread | about 4 ms, about 9 ms |
| Subscribe members one at a time (`member add`, no batching) | about 1.9 ms each at the store, plus one HTTP request |
Sending is bounded by the SMTP round trips to your relay: raise `relay.concurrency` for big lists.
## Licence
Parsimony is free software, released under the GNU General Public License, version 3 or (at your option) any later