# mvirt 0.24.6

migrations: no
wire: unchanged
rev: b675e83a24cf33bbd87a773c07bc805001175cd8

mvirt 0.24.6

A patch release: faster recovery after a host reboot, and no more cleartext
API on :80.

What changed

- #359: the crash-loop backoff counts failures since the container's last
  healthy run (10 min Running) instead of over its whole life. An app that
  fails once after a host reboot — typically because its database on the
  same host is still booting — restarts immediately instead of waiting out
  the 5-minute cap on every attempt (~20 min of 503 on the 0.24.5 roll).
- #194: the control-plane edge answers `http://api.<domain>` and the console
  with a 301 to https:// once a certificate covers the name; the ACME HTTP-01
  challenge keeps answering on :80, and the object gateway keeps plaintext
  (S3 clients do not follow redirects on signed requests). Tenant app
  ingress is unchanged (#286).

Upgrade: ordinary host-by-host roll (release.md §C1 + §C3). No store migration.


## Upgrading

Point your flake's `mvirt` input at this release and redeploy the control-plane
host first, then each node in turn:

```
nix run .#colmena -- apply boot --reboot --on <control-plane host>
nix run .#colmena -- apply boot --reboot --on <node>   # one at a time
```

This release adds no migrations, so the control-plane host rolls back the same way
a node does: boot the previous NixOS generation.