The inbuxa-server image (both 2026.9.23 and 2026.9.30.2) declares VOLUME ["/etc/inbuxa", "/var/lib/inbuxa"] and starts with --config /etc/inbuxa/config.json. So the configuration and all mail land in anonymous volumes, and inbuxa-etc / inbuxa-data stay empty.
Seen on a fresh --local install in the debian13 lab (2026-10-05):
docker compose up -d recreating the server keeps the anonymous volumes, so upgrades and restarts have worked.
docker compose down followed by up, or docker compose rm, gives the server new, empty anonymous volumes: it comes back in bootstrap mode with no configuration and no mail.
Backups of the named volumes contain nothing.
The fix isn't only the two mount targets. On an existing install, pointing the named volumes at /etc/inbuxa and /var/lib/inbuxa would lay the empty named volumes over the server's real data. It needs a migration step, run once by the installer on converge: copy each anonymous volume's contents into its named volume while the server is stopped, then recreate with the corrected mounts and check the server comes back configured. Add an e2e case for it (install with the old template, upgrade, check the domain and first mailbox still exist), and one that down then up keeps everything.
The compose template mounts the server's named volumes where the server never looks:
```yaml
volumes:
- inbuxa-etc:/opt/stalwart/etc
- inbuxa-data:/opt/stalwart/data
```
The inbuxa-server image (both 2026.9.23 and 2026.9.30.2) declares `VOLUME ["/etc/inbuxa", "/var/lib/inbuxa"]` and starts with `--config /etc/inbuxa/config.json`. So the configuration and all mail land in **anonymous** volumes, and `inbuxa-etc` / `inbuxa-data` stay empty.
Seen on a fresh `--local` install in the debian13 lab (2026-10-05):
```
volume inbuxa_inbuxa-etc -> /opt/stalwart/etc
volume inbuxa_inbuxa-data -> /opt/stalwart/data
volume 906a666e… -> /etc/inbuxa
volume f63a0665… -> /var/lib/inbuxa
```
What that means:
- `docker compose up -d` recreating the server keeps the anonymous volumes, so upgrades and restarts have worked.
- `docker compose down` followed by `up`, or `docker compose rm`, gives the server new, empty anonymous volumes: it comes back in bootstrap mode with no configuration and no mail.
- Backups of the named volumes contain nothing.
The fix isn't only the two mount targets. On an existing install, pointing the named volumes at `/etc/inbuxa` and `/var/lib/inbuxa` would lay the empty named volumes over the server's real data. It needs a migration step, run once by the installer on converge: copy each anonymous volume's contents into its named volume while the server is stopped, then recreate with the corrected mounts and check the server comes back configured. Add an e2e case for it (install with the old template, upgrade, check the domain and first mailbox still exist), and one that `down` then `up` keeps everything.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
The compose template mounts the server's named volumes where the server never looks:
The inbuxa-server image (both 2026.9.23 and 2026.9.30.2) declares
VOLUME ["/etc/inbuxa", "/var/lib/inbuxa"]and starts with--config /etc/inbuxa/config.json. So the configuration and all mail land in anonymous volumes, andinbuxa-etc/inbuxa-datastay empty.Seen on a fresh
--localinstall in the debian13 lab (2026-10-05):What that means:
docker compose up -drecreating the server keeps the anonymous volumes, so upgrades and restarts have worked.docker compose downfollowed byup, ordocker compose rm, gives the server new, empty anonymous volumes: it comes back in bootstrap mode with no configuration and no mail.The fix isn't only the two mount targets. On an existing install, pointing the named volumes at
/etc/inbuxaand/var/lib/inbuxawould lay the empty named volumes over the server's real data. It needs a migration step, run once by the installer on converge: copy each anonymous volume's contents into its named volume while the server is stopped, then recreate with the corrected mounts and check the server comes back configured. Add an e2e case for it (install with the old template, upgrade, check the domain and first mailbox still exist), and one thatdownthenupkeeps everything.