Hello @pathab Thanks for reporting. We will look into it.

Hello @firmansi I still need to look into this.

@girish said: We are looking into this again @locutus , thanks for reporting. Meanwhile, I have revoked the release. Works. Thanks.

Found the issue - https://git.cloudron.io/platform/box/-/commit/03e0e98d1a14140c469ef915f01638ca515d7cb8 . The platform code was trampling permissions of the build cache Well, lesson learnt. I have put tests in place now so that this doesn't happen again.

@james These days i use Cloudflare Zero trust to protect certain domains from public access, for example like in wordpress, I protect the access to wp admin with cloudflare, not some kind of vpn that need another private layer to connect with (as far as I know that how VPN Works) I have been using N8N for years, and frankly never had once the webhook of N8N exposed to public on front end code, always use only in backend. What you explain is really nice , but to my understanding of VPNs is that they require a client for access and introduce overhead, which can reduce overall connection speed I think a proxy firewall solution is more interesting, like what Cloudflare does. But when I first saw the article about using another domain for webhooks, I thought it was similar to Typebot's setup, where the console and the chat window use different domains.

@james somehow, it's deemed as a preview, not a stable one, but the documentation can be found here https://docs.n8n.io/build/ways-of-building-workflows/ai-assistant

@alwynispat SQLite is not as brittle as you think.

Hello @jkammerer It should be working.

Hi @james, thank you very much for the quick implementation and for adding Queue Mode support to the Cloudron n8n package. I have tested it now, and the difference is very noticeable. The n8n UI stays stable and responsive while larger executions are running, and the workload can finally make use of multiple CPU cores through the worker processes. This is exactly the behavior I was hoping for. In my case, it makes larger workflows much more usable and stable compared to running everything in a single n8n process. Especially with workflows that process large amounts of data, Queue Mode helps a lot because the main process no longer gets blocked by all the execution work. So far, the setup works well for me. Being able to configure the number of workers through the Cloudron package is very useful, and I think this is a great improvement for people running heavier n8n workloads on Cloudron. Thanks again Cloudron team for the fast response and implementation!

Hello @chrishvd I was not able to reproduce your issue. Please contact support@cloudron.io so we can look deeper into this.

I didn't realize! Thank you!

@zigmasdirigeant said: Btw, this shouldn't be inside N8N. This is more a request for an app, not N8N alone. I agree, I'm not even sure how I did that, I tried changing it but couldn't find how.

@umnz same thing happened with me i updated and n8n when into broken state i restarted and it worked

Hello @bedrijfstak14 Great to read that the Cloudron update 9.1.7 resolved this issue.

the update is live, thanks!

I did a clean install of n8n and it installed quickly. So something is/was up with my old n8n instance. Not a huge problem, I'll just migrate my workflows to the new and archive the old instance.

Hello @ekevu123 Disabling is not actually disabling, but more disabling a custom override of the default which is 300 seconds.

Glad to hear it! Thanks @nebulon.