Thanks for picking this up @james
lucidfox
Posts
-
HTTP/1.1 subscriptions never receive data (nginx proxy buffering) -
HTTP/1.1 subscriptions never receive data (nginx proxy buffering)Subscribing to a topic as a stream (GET /<topic>/json?poll=false, also /sse and /raw) never returns anything over HTTP/1.1 on my ntfy instance: the TLS handshake and request complete, but nothing comes back (not even the response headers). The same request over HTTP/2 works and delivers the open event and a keepalive every 45 seconds. Short requests over HTTP/1.1 are fine (poll=1, /v1/health, publishing).
This is for a private ntfy topic, and you can reproduce the issue like this:
curl -sN --http1.1 -m 120 -H "Authorization: Bearer $TOKEN" "https://ntfy.example.com/mytopic/json?poll=false"
-> nothing for 120 seconds
curl -sN --http2 -m 120 -H "Authorization: Bearer $TOKEN" "https://ntfy.example.com/mytopic/json?poll=false"
-> {"event":"open",...} at once, then a keepalive every 45 sntfy's reverse-proxy docs (https://docs.ntfy.sh/config/#nginxapache2caddy) require
proxy_buffering off(andproxy_request_buffering off) for this reason. Clients that speak HTTP/1.1, such as Python's httpx or the ntfy CLI in some setups, hit this; browsers and curl usually negotiate HTTP/2 and don't.Would it be possible for the ntfy package's nginx location to get
proxy_buffering off? -
🔥 MIROTALK ULTIMATE BUNDLE — LIMITED-TIME SALE — ENDS 23/01/2026! 🔥 -
Nextcloud update pushed too early?just fyi, an update for team folders to 20.1.1 has been released and fixed the group folders issue for me.
-
Nextcloud update pushed too early?There seems to be a working patch for version 20.1.0 of group folders - https://github.com/nextcloud/groupfolders/issues/4037#issuecomment-3352521147
is this possible for us to implement on cloudron?
-
Adding MCP in LibrechatI've been trying to get Librechat setup (thanks for getting the package on Cloudron), and have got Openrouter and websearch working (using a SearXNG hosted on Cloudron). One of the features that would dramatically increase the capabilities of the LLMs on Librechat would be to leverage the Model Context Protocol (MCP) and integrate them with various tools.
According to the Librechat documentation on MCP - [https://www.librechat.ai/docs/features/mcp] - the easy way to find and install MCP servers on Librechat is to use [smithery.ai]. But executing an npx command in the Librechat app terminal on Cloudron doesn't work. I guess this is because the file system is read-only? Or the folder structure is not what the smitthery.ai code expects? Or both?
I also tried to manually add an MCP server - for Exa Search - using this in the librechat.yaml file:
mcpServers: exa-search: command: "npx" args: ["-y", "@exa-labs/exa-mcp"] env: EXA_API_KEY: "${EXA_API_KEY}"It shows up on the Librechat interface, but is 'offline' and cannot be initialized.
Has anyone had some luck getting an MCP going on Librechat? Or any tips on what might get it working?
-
Running 2 instances of Syncthing on 1 server -
SendGrid is over, what to use instead?I've been using the hosted version of plunk - https://www.useplunk.com/ - which is an open source email relay/transactional email service app built on top of AWS. They've got a generous free tier and the paid plans are very reasonable pay as you go.
-
SolidtimeSolidtime is a modern open-source time tracking application for Freelancers and Agencies.
It could be a really useful addition to Cloudron
AGPL 3.0
I've played around with it a bit, the UI is quite slick and it seems fairly feature complete. Potentially a great alternative to Toggl, Clockify etc
-
Notifications not working? -
Notifications not working?I'm facing the same issue that @odie described. Test Ntfy notifications are working properly, but not any of the actual subscription notifications.
When I try the /app/pkg/cron.sh sendnotifications command I only get this response -
2025-04-02 16:06:56<br />
I can see this repeatedely in the logs as well -
Run cron job 'sendresetpasswordemails'
Run cron job 'sendverificationemails'Has the issue been resolved for anyone else or am I missing something?
Also, in case it matters, none of the subscription notifications I've setup are from the admin user account.
-
My Hetzner Storage Box Volumes and Backup locations needed remounting post upgrade to 8.0.3 -
My Hetzner Storage Box Volumes and Backup locations needed remounting post upgrade to 8.0.3 -
My Hetzner Storage Box Volumes and Backup locations needed remounting post upgrade to 8.0.3 -
Recommendations for Self-Hosted Collaborative Email ClientsI'm not sure if they have all the features that you're interested in but Chatwoot and Freescout on Cloudron might fit the bill.
I've used Missive quite a bit, it's not open source or self-hosted, but is a pretty good service.
-
Synapse and Peer Tube Not Responding -
Synapse and Peer Tube Not Responding@girish thanks for the response.
the
server_namefield was missing from the config file for some reason. i added the field at the top of the file and restarted the app, but am still getting these errors in the matrix app log:Healtheck error: Error: connect ECONNREFUSED 172.18.19.36:8008OSError: [Errno 30] Read-only file system: '/app/code/media_store'the peer tube app does have a
/app/data/storage/logs/directory with log files in it. this app is still not responding as well. i can see this in the error logs:errno: -2, -
Synapse and Peer Tube Not RespondingThe Synapse and PeerTube apps have stopped working on our Cloudron. This happened after a reboot for system updates, and it's a bit of mystery why they should suddenly get misconfigured. I've tried several restarts and restoring from a backup, but it hasn't fixed the issue. Renewing Certs and DNS hasn't helped either.
The logs for the Matrix/Synapase show these errors:
Healtheck error: Error: connect ECONNREFUSED 172.18.19.36:8008=> Starting synapse
Error in configuration:
Missing mandatoryserver_nameconfig option.The logs for PeerTube shows these errors:
Healtheck error: Error: connect ECONNREFUSED 172.18.16.78:90002022-10-16T08:21:42.000Z ==> Starting PeerTube
throw err;
^
ENOENT: no such file or directory, mkdir '/app/code/server/storage/logs'Any help to troubleshoot and get these apps back running would be much appreciated.
-
Mobile app -
Mobile appThanks for packaging this! It looks like it already is/and could become a great photo app. I was just trying to get an andoird mobile app connected, but it doesn't seem to be able to login. The server endpoint I'm using is http://[cloudron-app-location]:2283/api
Has anyone had any luck yet with using the Immich mobile app?
. The notifications are firing as expected now.