Releases
Changelog
Published versions. Cluster and Fleet are separate tarballs. Sign in to download.
Fleet fleet-0.1.0-rc.1
Sep 1, 2026
lattice-fleet 0.1.0-rc.1 — 2026-09-01
Release-candidate Fleet tarball (control plane + agent). Not the DHCP/IPAM cluster package. Still pre-GA — not general availability.
This archive is signed (
SHA256SUMS.asc). Install with systemd.- Fleet is a manager of managers (HQ console). If this VM is down, DHCP at the sites keeps working.
- First Fleet RC. Two Fleet VMs behind the included HAProxy example, a shared Postgres with SSL (
sslmode=requireorverify-full), and live cluster agents were proven on Ubuntu VMs on one LAN: enroll a paid HA site, fail the site VIP with DHCP still working, fail the VM that runs the agent and enroll on another node, enroll a second (community) site, stop one Fleet node (wait a few seconds for the load balancer), rolling upgrade, restart the agent without a new token. - Agent on a cluster VM. One agent per site, on one control-plane VM. Installing the agent does not change permissions on that machine’s Lattice files (that broke paid HA on 0.1.0-beta.18 / 0.1.0-beta.17). If that VM is down, DHCP at the site still works; enroll the agent on another node when you want the console back.
- Replace an agent.
fleet-install --mode agent --forcewith a new enroll token starts a new identity. A leftover saved key from a previous enroll is discarded. - HA. Same sign-in token and encryption key on both Fleet VMs (
--credential-key). If one Fleet VM stops, wait a few seconds; cluster pages in the console come back. Sites keep leasing. - Needs Lattice cluster 0.1.0-rc.2 or newer on paid HA sites (single-node cluster 0.1.0-beta.17 still works).
- Upgrade is still
fleet-install --mode upgradefrom this archive. Two Fleet VMs: one node, then the other. If you already run 0.1.0-beta.19, this is the release-candidate stamp (same installer). If you run 0.1.0-beta.18, this also includes the agent-on-cluster and re-enroll fixes.
Lattice 0.1.0-rc.3
Sep 3, 2026
lattice 0.1.0-rc.3 — 2026-09-03
Release-candidate cluster tarball. Still pre-GA — not general availability. Same binaries for Community and paid; the signed entitlement is the SKU.
This archive is signed (
SHA256SUMS.asc). It includes the installer and EULA.- DHCP: two servers on the same cluster look for free addresses from opposite ends of the pool. That is a search hint, not a split — either node can still grant any free address. A Discover does not hold an address until the client Requests it, so a client that talked to one node can Request on the other.
- A burst of new devices is less likely to stall DHCP or take the VIP health-check down. The license cap is counted when a lease is actually granted, not when an offer is sketched.
- HA join and relayed DHCP from 0.1.0-rc.2 are in this archive (join no longer rewrites the first node’s database passwords; trusted relays match the helper’s VLAN address even if Docker rewrote the packet source).
- Three Ubuntu 22.04 or 24.04 VMs on one LAN: install the primary, join two more, kill the VIP holder with DHCP still working, upgrade replicas then the leader, backup and restore onto a new cluster. Still pre-GA.
- Upgrade is still
lattice-install --mode upgradefrom this archive. Three-node HA: replica, replica, then the current leader. If you already run 0.1.0-rc.2, this is the DHCP-grant cut. Community single-node is unchanged except DHCP.
Sign in or create an account to download.