Cloudron Forum

We recently made some updates to the windmill build here:

https://github.com/halecraft/windmill-cloudron

Several patches now enable or re-implement features that artificially cap the open source code--we build Windmill from its AGPLv3 source and wire the open-source stubs to the open-source implementations that upstream ships but leaves disconnected.

Running ./console core:update in the Matomo app's terminal updated the DB for us.

Since upgrading to the latest Matomo, we are now seeing the following error inside the "Visits over time" widget within the app:

SQLSTATE[42S22]: Column not found: 1054 Unknown column 'log_visit.ai_agent_name' in 'where clause'

What's the best way to address this?

@jdaviescoates said in Unified messaging platforms in 2026 (beeper, matrix-mautrix, ?):

@canadaduane I think @andreasdueren has packed WhatsApp and Signal bridges for Matrix

I think I found it! Thank you.

https://forum.cloudron.io/topic/14906/matrix-signal-bridge

@drm237 said in Windmill: Open-source alternative to Airplane, Superblocks, Retool:

I submitted a pull request if anyone wants to give it a try: https://github.com/halecraft/windmill-cloudron/pull/1. It assumes that @canadaduane didn't update the init-db-as-superuser.sql file in their original build and instead fetches the original file from windmill.

Merged! And I'm sorry it took so long to see it. I must have cleared out my github inbox without noticing.

@scooke said in Unified messaging platforms in 2026 (beeper, matrix-mautrix, ?):

I like keeping them all separate; different social circles use different platforms, and I like the gap between them all. Why would I want to have one source where I see all 14 messaging apps with all those messages???

Totally fair! My use case is less about apps and more about wanting to reclaim control over my digital life. I'd like to have an AI assistant (of my own making) that can (based on configurable rules and heuristics) help raise and lower the priorities of some messages and classes of messages (e.g. advertising, status updates from airlines, status updates about packages, etc.) Google has kind of achieved this with its email inboxes and smart patterns. But I don't want to be dependent on Google to be aligned with my best interest.

So far I'm considering matrix-mautrix as a nice middle ground for this, but I haven't yet built a cloudron package for it, so I'm not sure how difficult it might be.

Duane

@sangemaru said in MCP Server for Cloudron - AI-Powered Instance Management:

Would be great if you can make your fork repo public.

Done!

https://github.com/halecraft/mcp-cloudron

@sangemaru said in MCP Server for Cloudron - AI-Powered Instance Management:

And, I hope you'll excuse the pun, but you could totally rock "The Rock" Johnson as a nickname.

Ha! I have been told my nickname should be The Pebble (if you compare, you would see why :D)

Have a great day,

Duane

With apologies in advance to @sangemaru (since I did not ask permission), I've forked this project.

• the git repository's source code is in a poor state
  • node_modules checked into source code
  • dist checked into source code
  • multiple megabytes of executables and directories not compatible across operating systems
  • many leftover markdown files, plans, unused files throughout
  • tests not working, and typescript unused
• Since I used a filter-repo to clean up the git history, this made my changes incompatible with github PRs

Subset of filter-repo commands used during cleanup:

git filter-repo --force --path node_modules --invert-paths
git filter-repo --force --path dist --invert-paths
git filter-repo --force --path TESTING.md --invert-paths
git filter-repo --force --path PHASE_2_3_SUMMARY.md --invert-paths
git filter-repo --force --path .serena --invert-paths
git filter-repo --force --path FORUM_POST.md --invert-paths
git filter-repo --force --path REAL_TEST_RESULTS_F23B.md
git filter-repo --force --path TESTING_NOTES.md --invert-paths

Here is the new published package:

https://www.npmjs.com/package/@halecraft/mcp-cloudron

I separated the API surface area into 8 categories for now, and implemented a few more of the openapi spec endpoints (32 of 240 total implemented):

App Management

• cloudron_list_apps - List all installed applications with status and health
• cloudron_get_app - Get detailed information about a specific application
• cloudron_control_app - Start, stop, or restart an application
• cloudron_configure_app - Update environment variables, memory limits, access control
• cloudron_install_app - Install an application from the App Store
• cloudron_uninstall_app - Uninstall an application (with pre-flight validation)
• cloudron_clone_app - Clone an application to a new location
• cloudron_repair_app - Repair a broken application
• cloudron_update_app - Update an application to a newer version
• cloudron_restore_app - Restore an application from a backup
• cloudron_backup_app - Create a backup of a specific application

App Store

• cloudron_search_apps - Search the Cloudron App Store
• cloudron_validate_manifest - Validate an app before installation

Backups

• cloudron_list_backups - List all available backups
• cloudron_create_backup - Create a new instance backup

Users

• cloudron_list_users - List all users
• cloudron_get_user - Get details for a specific user
• cloudron_create_user - Create a new user with role assignment
• cloudron_update_user - Update user properties (email, role, password)
• cloudron_delete_user - Delete a user (with pre-flight validation)

Groups

• cloudron_list_groups - List all groups
• cloudron_create_group - Create a new group

System

• cloudron_get_status - Get Cloudron instance status and version
• cloudron_check_storage - Check available disk space
• cloudron_list_services - List platform services (MySQL, PostgreSQL, etc.)
• cloudron_list_domains - List configured domains
• cloudron_get_logs - Retrieve application or service logs

Tasks

• cloudron_task_status - Check status of an async operation
• cloudron_cancel_task - Cancel a running operation
• cloudron_validate_operation - Pre-flight validation for destructive operations

Updates

• cloudron_check_updates - Check for Cloudron platform updates
• cloudron_apply_update - Apply a platform update

I don't normally do this (I like to contribute), but I have some time constraints right now and am currently unwilling to slow down to work through some of these issues to get to baseline. That said, I do applaud your getting this started sangemaru and hope this doesn't dissuade you from continuing.

I have an app configured incorrectly, and I occasionally notice this message in my Cloudron console:

gaiacircle.org

    PTR4 DNS record (PTR) did not match.
        Hostname: x.y.z.w
        Expected: [my.cloudron.instance]
        Actual: x-y-z-w.googlefiber.net

See troubleshooting (link) docs for more information.

The "troubleshooting" word is a link, and it points here:

https://docs.cloudron.io/troubleshooting/#mail-dns

However, as far as I can tell, there is no "#mail-dns" anchor on that page.

Does the link (or the doc) need to be updated?

What's the current status of unified messaging platforms in 2026?

I see a few threads from 3 years ago regarding beeper when it was making news.

I would love to be able to put each of us at the center of our lives--using communication platforms as bridges, rather than the way they wish to be used (as islands of corporate ownership and monetization of user attention and influence).

Is matrix-mautrix good? Does anyone use it in conjunction with the cloudron matrix instance?

Is beeper's self-hosted option good?

Are there alternatives to consider?

Thanks!

The bitwarden desktop client app (2025.12.0) now requres an endpoint at /api/tasks when editing an entry. If that endpoint is not available, the client will report "An unexpected error has occurred", or something about an incorrect cipher.

The actual cause of the error is reported here: https://github.com/dani-garcia/vaultwarden/pull/6557

Until Vaultwarden is updated, users will continue to experience this with an upgraded desktop client.

I had this happen at a recent upgrade to Cloudron 9. All apps but Vaultwarden were working, and the "Cloudron has no app configured for this domain" message was showing. Indeed, it was just a hard refresh to solve it in Zen / Firefox (how odd!)

In case this is useful to anyone else:

I wanted to gain access to the full meetings and token APIs at my-server.com/api/v1/docs/:

Since I was able to access the page above, it looked at first like everything was already set and ready to go!

However, I soon learned that this was not the case:

{"error":"This endpoint has been disabled. Please contact the administrator for further information."}

This made me think that I needed to use the secret in /app/data/api_secret to access the API. But it didn't work (same error message as above).

It turns out I needed to add the following to the /app/data/env file:

API_ALLOW_MEETINGS=true
API_ALLOW_TOKEN=true

Once I restarted the server, I was rewarded with a new error!

{"error":"Unauthorized!"}

This unauthorized error shows up because, I believe, the /app/data/api_secret is not used for the meetings or token APIs. You need to ALSO add the following to your /app/data/env file:

API_KEY_SECRET=yourVerySpecialSecret

Finally, you can click the Authorize button in the swagger / MiroTalk SFU API page (shown above), and add yourVerySpecialSecret to the input box there. After that, you have access to your own MiroTalk SFU API!

Update to add: You can find a complete list of possible env settings here:

https://github.com/miroslavpejic85/mirotalksfu/blob/main/.env.template

Hi nebulon! Thanks for your thoughtful response. Some notes below.

@nebulon said in Show app disk usage in Storage tab:

So this is either not trivial or certainly misleading. Either this will show a size much larger if you would add up all the apps (if layers are shared) or if we would somehow count shared layers only towards one installed app, would that one be uninstalled, suddenly another app would take up more space.

My particular need would be resolved by the phrase "up to X GB" for the base image, e.g. "up to 1.2 GB" just to give me a rough estimate of things.

@nebulon said in Show app disk usage in Storage tab:

The system info page shows the data used by the app as such already. This is also much improved in the next Cloudron version.

The system info page does NOT show what I am looking for. In particular, for apps that are mounted on a drive other than the default ubuntu image, I do not see the size of the data I'm using. For example, I have Open WebUI at "ask.halecraft.org" but this is nowhere visible on the stats page:

I can't imagine I'm the first to ask for this, but I also can't seem to find any previous discussion on this. The docs talk about setting mounts and moving data, but do not discuss "how much data" I'm using nor how to find it.

I want to know how much space an app is using--both:

• the "base" disk usage (i.e. docker image, and whatever necessary system files it provides), and
• the "user" disk usage (i.e. the additional data on volume mounts or whatever I've configured the app to use to store as it functions)

I realize this might be a costly operation to discover--but a button in the Storage tab with "Measure Current Disk Usage" or something like this would be an acceptable trade-off.

Am I missing an obvious area of the app UI where this is already shown?

Oh wow, what an interesting (and slightly frightening) failure mode! "Your log files are so big that we've given up on managing them."

@msbt said in Windmill: Open-source alternative to Airplane, Superblocks, Retool:

@canadaduane cool, but it seems it's not yet available to the public

Oops! My MIT-licensed code is now public

We have a working Windmill instance on Cloudron, (container) released with MIT license here:

https://github.com/halecraft/windmill-cloudron

The compromise we had to make here was to embed a Postgresql server inside the container, rather than connect to the one provided by Cloudron's postgres addon. Short of modifying Windmill itself, this was the only way we could give Windmill the full control over the Postgres instance that it needs / assumes it has.

@girish Sorry, looks like I got it confused with Windmill. I can't find the Kestra demo.

I've been dabbling with trying to create a Windmill on Cloudron container. It's more difficult than other apps. Some notes:

• There is a hard-coded set of ROLEs required of the postgres database: windmill_admin, and windmill_user. Windmill uses these in DB migration scripts as well as during runtime to control access and access levels (see this section). As far as I can tell, Cloudron's postgres does not offer flexibility to add new roles in the database.
• Windmill's LSP (Language Server Protocol) is itself a docker container with various programming languages (node, deno, python, etc.) pinned to certain versions. Because of these pinned versions, it is difficult to integrate into a single container along with the rest of Windmill. This suggests the docker Cloudron addon may be necessary.
• Windmill also runs workers within docker containers. This may also require the docker Cloudron addon.

In my effort to get most of Windmill running inside the same container, I have supervisord running with the following processes:

• caddy
• windmill-server
• windmill-worker

Caddy (or similar) is needed in order to act as reverse proxy for both windmill-server and the windmill-lsp container's websocket service.

The biggest issue at this point is the hard-coded windmill_admin and windmill_user postgres roles. I'm not sure how to get around this other than to create a separate postgres DB, unfortunately.