# mvirt 0.24.10

migrations: no
wire: unchanged
rev: 6e554c62c11a0993c80c8b50df9597b2e1c4d522

mvirt 0.24.10

A patch release: diagnostics for the slow first slice push after an election,
and a readiness fix for the gateway agent.

What changed

- #341 (diagnostics): every top-level step of the converger ladder is timed
  and WARNs with its name above 5 s, and a whole pass above 10 s WARNs with
  whether it held a leadership's first push. On the 0.24.9 roll a fresh
  leader still pushed its first slices 7 min after the election with no log
  line in between; this names the step.
- #293: the gateway agent offers its `:19001 /ready` listener only while it
  serves what its document declares — not when an HTTPS listener and
  certificates are declared but no private key could be fetched (a restart
  with the control plane unreachable), so such a member leaves the Maglev
  pool instead of resetting a third of the TLS flows. Shipped here for the
  control-plane host edge; the tenant fleet picks it up with the next
  gw-agent image.

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.