Release state
Replaced #
This was the Latest release until 2026.10.9 replaced it. Before it became Latest, we updated our own Moorhold server to it from 2026.10.7, and it passed a restore drill there. Its files stay on get.moorhold.dev and still work. We haven't tested an update from this release to 2026.11.4. If your server runs it, email hello@moorhold.dev and we'll help you move it.
We wrote this release's notes, including their Still to be tested line, before we ran it on our own server.
This release checks a server before the install asks for your access key, and makes Moorhold stop cleanly when the server shuts down, so its services start in order after a reboot. The launchpad now shows which release you run. Airflow asks you to sign in again once your Moorhold session has timed out, and DataHub's search leaves out the system tables.
https://moorhold.dev/releases/2026.10.8Before you update #
- What changes
- A fresh install now checks the server before it asks for your access key, and the launchpad shows which release you run. Moorhold now stops cleanly when the server shuts down, so its services start in order after a reboot.
- Downtime
- Moorhold is unavailable from when the update stops the running release until the new one has started. That hasn't been measured on this release yet.
- Backup
- Yes: before it changes anything, the update takes a rollback point, a backup of every database on this server. If the new release doesn't start, it goes back to that point by itself.
- Manual steps
- None for the update. A server updated to 2026.10.6 that restarts before it has this release needs sudo moorhold start --profile <profile> after the reboot.
- Still to be tested
- Approving an update in the portal, the portal before a Stack is applied or with no agents policy, and Airflow's scheduled runs and API tokens are still to be tested. So are a lake restore while a service is slow to stop, the LVM and sleep notes, and the /setup hint after a failed install.
Before you reboot a server on 2026.10.6 #
A server updated to 2026.10.6 from an earlier release may not start Moorhold again by itself after a reboot. If it restarts before you've updated it to this release, run this once it's back up:
sudo moorhold start --profile <profile>
Use the profile the server runs, such as full. Updating to this release fixes it for good.
Updating #
To update a server running 2026.10.6 or 2026.10.7, download this release's installer and run it on the server as you did to install, but without --hostname:
curl -fsSLo install.sh https://get.moorhold.dev/2026.10.8/install.sh sudo bash install.sh
The installer sees the release that's running and updates it in place, with the same checks and the same rollback point as before. On a server running 2026.10.7, admins can also ask for the update and approve it on the Updates page in the portal.
Every update now also sets the server to stop Moorhold cleanly when it shuts down, starting with the first reboot after the update.
New #
- Checks before an install. A fresh install now checks the server before it asks for your access key or installs Docker. It checks the operating system, memory and free disk for the profile, ports 80 and 443, and the network. Anything that would stop Moorhold working, such as nginx on port 443, stops the install there, with nothing pulled. Network findings are warnings, and behind a router without NAT loopback the check recommends the optional relay, on the same hostname. On a desktop install it warns if the server can still go to sleep, and when an LVM volume group has space that no filesystem uses yet, it gives the command to add it.
- Check a server without installing.
sudo bash install.sh --check --hostname <name>runs the same checks and stops. It asks for no key and installs nothing. If the server lacks a tool the check needs to verify the release, it names the package to install. On a server that runs Moorhold,sudo moorhold checkruns the checks too. - Clearer install output. Run in a terminal, a fresh install asks which profile you want and whether to load the sample data, when the flags don't say. It shows its stages under headings and counts the images as it pulls them. Docker's own lines and the
warning:lines meant for developers no longer show. It ends with the command to read your setup code and the address to use it at. - Moving to another hostname. Running the installer again with a different hostname now warns that every passkey stops working at the new hostname, and asks before it changes anything. Without a terminal it changes nothing unless you add
--change-hostname. - Your release on the launchpad. The launchpad now ends with the release of Moorhold the server runs, linked to that release's notes on moorhold.dev.
- Confirm an update on its own screen. From this release on, "Review the update" on the Updates page opens a screen that shows the release you're updating from and to, and what the update changes. If you're the only admin, it shows your current version and the new one right above the box, and you type the new version to confirm. With two or more admins, you ask another admin to approve, and they see the warnings you saw, even if a later check has found another release. While the update runs, the page says the portal restarts. After a failed update, it says the server is running the earlier release again.
- An update started on the server, approved in the portal. From this release on, an update you start from the server's command line that needs a second admin waits on the Updates page. One admin asks for it there and another approves it. This works for updates from 2026.10.8 to a later release.
- Drop the databases a rollback keeps. A rollback keeps each database it replaces under a dated name, and those copies are backed up like any other database until you drop them.
sudo moorhold backup drop-keptlists them with their sizes, and drops them once you say yes. It never drops a copy whose own database isn't beside it.sudo moorhold statuslists the kept copies, and a rollback's last lines say how to drop them. After a rollback to an earlier release, which doesn't have the command, those lines give the command to drop them by hand. sudo moorhold stop. Stops Moorhold's services and keeps their data, as the server now does at shutdown.sudo moorhold start, or the next boot, starts them again.- The agents policy on the Stack page. The Stack page has a new Agents section that shows who can propose changes to the stack and who approves them. When there's no policy yet, it says why.
- DataHub without system tables. DataHub's search and browse now leave out the tables Trino and Postgres keep about themselves, such as
information_schema, which showed beside your datasets, and so does the portal's search. DataHub does this with a view, "Datasets without system tables", which Moorhold makes the default when DataHub has none. Pick no view to see every table. Nothing is deleted.
Fixed #
- Airflow after your session times out. This is a security fix. Airflow now asks you to sign in again once your Moorhold session has timed out. Before, it could go on showing you signed in. If you're still signed in to Moorhold elsewhere, Airflow signs you in again by itself and comes back to the same page, but that reloads it, so anything you'd typed into a form and not saved is lost.
- Services at boot. After a reboot, Docker started many of Moorhold's services before Moorhold had put their secrets in place, so they failed and restarted until Moorhold started. Moorhold now stops its services itself when the server shuts down, so after a reboot they wait for Moorhold to start them. A service that stops by itself while the server runs is still restarted.
- A new look for the sign-in pages shows at once. Browsers now check for a new style each time they load a sign-in page, so a release that changes it shows straight away. Before, a browser that had seen the old style could keep it.
- The portal before the Stack is applied. Every page of the portal failed with a server error while the Stack wasn't applied yet, or had no agents policy. Pages now open, and the Stack page says what's missing. Single sign-on waits for the policy, and the local admin account works until then.
- Alarming lines in a healthy install. A healthy fresh install no longer says a service is unhealthy or that
docker compose upfailed. The portal's MCP server, which waits for the Stack, now starts once the Stack is applied. - The setup code after a failed install. When an install stops after it has made your setup code, it now ends with the command to read the code and the address to use it at. The setup page's hint gives the same command, for a terminal install and a cloud-init one.
moorholdwithout sudo. As a normal user,moorholdnow says to run it with sudo and shows the command to type, where before it failed with "Permission denied". On a server where it can run,moorhold statuswithout sudo says that backups and services need sudo. Before, it printed Docker's error.- Blank lines in an access-key file. Moorhold refused a valid key when its file had extra blank lines or spaces around the key. It now ignores them everywhere it reads a key file. A key with a space inside it is still refused.
- Restoring a backup while a service is slow to stop.
sudo moorhold backup restoreandsudo moorhold backup --lakerefused when a service had to be stopped by force. They now say so and go on, as a rollback has since 2026.10.7. - "Last backup" before the daily backup has run.
sudo moorhold statussaid "Last backup: none yet." beside a rollback point and a passing drill. It now names the newer of the two, and says the daily backup hasn't run yet. - The first admin on the oldest servers. On a server first installed with 2026.10.0 or 2026.10.1, the setup page couldn't make the first admin's Keycloak user, because Moorhold's own Keycloak account lacked a role. The update now gives it that role, as a fresh install has had since 2026.10.2, and says what it did. A server that uses your organisation's own sign-in doesn't need the role and is left as it is.
- The portal on small and wide screens. On a phone, short pages such as Sources no longer leave a gap under the menu, and a table no longer makes the page scroll sideways. The drawings beside the portal's notes now stay inside them, and the missing-page screen's boat is big enough to see.
Verified #
- Unit and golden tests: pass.
- Images: every image passed the licence check.
- Tested on test servers: fresh installs on the core profile with the new checks; the updates from 2026.10.6 and 2026.10.7, each taking its rollback point; a rollback to 2026.10.7; and a reboot after each. Our own server on the full profile updated from 2026.10.7, and its restore drill passed.
- Owed: approving an update in the portal, the portal before a Stack is applied or with no agents policy, and Airflow's scheduled runs and API tokens. So are a lake restore while a service is slow to stop, the LVM and sleep notes, and the
/setuphint after a failed install.
Images #
Every one of Moorhold's own images is rebuilt from this release's source. No tool changes version, and the upstream images keep the digests they had in 2026.10.7.
Still known #
- Checking a server with
--check. On Ubuntu,sudo bash install.sh --checkstops with "Permission denied" after its downloads, because Ubuntu doesn't let programs run from the place the check unpacks them to. A fresh install still runs the same checks before it asks for your access key, andsudo moorhold checkworks on a server that runs Moorhold. Fixed in 2026.10.9. - Moving to another hostname: give the profile again. Run the installer with the new hostname and the profile the server was installed with, for example
sudo bash install.sh --hostname <new name> --profile core. Without--profile, the installer asks for the profile again on a terminal, and without a terminal it assumes the full profile, so on a 16 GB server it stops at the memory check. Nothing is changed either way. Fixed in 2026.10.9. - An install that never finishes. Rarely, the restore drill at the end of an install or update waits for ever on the sample data while it loads, so the installer never finishes. 2026.10.7 does this too. 2026.10.9 doesn't fix it yet. To finish the install, run this from a second terminal on the server:
sudo docker compose -f /opt/moorhold/project/compose.yaml exec -T postgres psql -X -U postgres -d moorhold_warehouse -c "SELECT pg_cancel_backend(pid) FROM pg_stat_activity WHERE usename = 'moorhold_loader' AND wait_event_type = 'Lock'"
The install then finishes. The data load stops with an error, keeps the data it had and runs again at its next scheduled time. We're working on a fix.
- DataHub after signing out elsewhere. DataHub can't be told when your Moorhold session ends. If you sign out of Superset or Airflow, DataHub stays signed in until its own session runs out. Signing out from the launchpad or from DataHub itself signs you out of DataHub too.
- DataHub on a shared computer. If the last person to use the browser didn't sign out, DataHub can still open as them, with their access, until its session runs out, even after you sign in to Moorhold as yourself. Sign out before someone else uses the browser.
- Users and Groups in DataHub. If your role can't manage users, DataHub's Users and Groups page shows "Failed to load users!". That's DataHub's message when your role isn't allowed to see the list of users.
- Working only in Airflow. Time spent in Airflow alone doesn't keep your Moorhold session going, so if you use only Airflow until the session times out, you see the sign-in page.
- API tokens made in Airflow. They keep working after you sign out, until they expire or are deleted in Airflow.
- System tables for agents. Agents that search through DataHub's own MCP server still see the system tables, because DataHub applies its default view only on its own pages.
- Docker's lines during an update. An update still shows Docker's own output while it pulls the new images.
- A reboot after a rollback. A release from before this one can't stop Moorhold at shutdown. So after a rollback to one, a reboot shows services failing and restarting until Moorhold has started, as before. They recover by themselves.
- Airflow's Triggerer. Airflow shows its Triggerer as unhealthy because Moorhold doesn't run one. Nothing in Moorhold uses deferrable tasks, and Moorhold's own health check ignores it.
- x86_64 (amd64) servers only.
Upgrading #
On a server running 2026.10.7 or later, update from the server's command line with sudo moorhold upgrade <version>, giving the release you're moving to. On a server running 2026.10.6 or later, you can instead download the new release's installer and run it on the server as you did to install, but without --hostname.
This release's notes give its steps under Updating, and say which releases it updates from. If they don't cover the release your server runs, email hello@moorhold.dev and we'll help. More about upgrading
Recovery steps #
There's no automatic rollback yet, so there are no recovery steps to follow on your own. If something has gone wrong with your Moorhold, email hello@moorhold.dev with the version you're on and what you saw, and we'll help. Please don't send passwords or keys.
Download and verify #
The signed release is on get.moorhold.dev. Check the signature on release.lock first, then each file's SHA-256.
| File | SHA-256 |
|---|---|
| moorhold-2026.10.8.tar.gz | fcb0aa8eeca45413c08324d010e3c7d1a0a2128e0e8e2332d0e2f13ff29bf9dc |
| manifest.yaml | 37150b51c5f945f6b3409c8ab8ee25a673eed64c5bf2c1aa4ec2541a138b07d7 |
| release.lock | a4f89267919d3c956b97ed38b1e81074c4a1f4bcb8fe089e04c3390ceeb7d542 |
| install.sh | 4ef7043c866fd9f67d50b40cd7d7fcb86b7b20f1beada3f7f140b3b1b7547ad4 |
| cloud-init-ionos.yaml | 17fa1928e0eed61a2d5e18ec5143bb28dd3278f5c19253d9dc6e28af817cd171 |
| cloud-init-hetzner.yaml | a2ebc05f9db897324876743bc9a58bb1b27ac1410f9b502f42ad2c2f9b13b052 |
| release.lock.minisig | Signature, key 7F6D12F3F6217F9C |
| SHA256SUMS | The checksums above |
minisign -Vm release.lock -p moorhold-release.pub sha256sum -c SHA256SUMS
Public key: get.moorhold.dev/moorhold-release.pub