Sponsoring: n8n
-
Where is the blocker right now? Is it the package functionality or just getting on the app store.
I remember there was an issue with web hooks working correctly but I'm not sure where that went.
-
@atrilahiji My understanding was there was concern with the lack of user management, but then the custom auth feature should solve that somewhat.
I also had a lengthy exchange with the project creator and main developer, who really wasn't mad about open-source or Cloudron at the time of the conversation, so I didn't know what to make of the intentions there.
The biggest blocker with anything is that there's a catch-22 for testing time, where many users will help test and give feedback on something but few have the wherewithall to get apps any other way than from the app store.
I expect because the majority have day-jobs that aren't dev-ops, so just like building a kit-car, it'd still be quicker and cheaper for me to do my day-job to buy a car than to build one. Which leaves us with about a dozen app-packaging benefactors running on goodwill, and a lot of patience and hope from everyone else in between.
For me ETL is the life-blood of modern business and life, and currently there are zero options for it in the App store. Whereas there's a dozen Kanban boards and salivation every time there's another app that uses the same styling as Notion, as if that somehow make it good.
Incidentally, I've sponsored things before and they just took the money and ran without finishing. I just don't know what to do with that, I'd rather cut off a finger than not keep a promise, so now I'm opting for wait and see, and focus on what I can do something about, and that's the business I generate using what we already have, which in this area is a recommendation to try Integromat instead, and leave the rest to hope and chance.
-
@marcusquinn Yeah fair enough. Did that happen in the cloudron community? O_O
Also, I was thinking of an "alternate" app store where you would need to enter your cloudron URL and an API key, and then select from a number of pre-built apps I can throw up. Of course with the intention that it is "one click" in a way, and things could be unstable.
Is there an appetite for this? I don't want to step on the toes of @staff but it just crossed my mind to make this for myself and friends but I can open it up. It very well could run into the limited person-power issue that the current app store has but just a thought.
-
@atrilahiji I like the idea. I just can't do much to help as my kind of work is more focused on the usage of apps and businesses reliant on them.
I just cannot stop the world from turning, so when n8n isn't available, I go to next-best alternatives like Integromat.
It's all very well having open-source preferences, but when there's no options within reach that will cover the cost of my time, I have to go with what works over what's hoped to maybe one day if someone else agrees might work and might stay working.
I can point at the best apps that are good for providing value to business, and therefore enabling earning for consultants working with CLoudron as an integral part of their toolkit.
What I can't do is help any business or person when interest gets diverted into entertainment and gaming apps. I respect they are nice - but they aren't solving the bottleneck that is stopping good quality, legitimate value generating apps from getting launched.
-
@atrilahiji I believe the main issue with the package that's available is that there's no connection to the server, so it won't actually run any workflow.
I do love your idea of a crowd-sourced App Store, of course with the staff's blessing as you mentioned. There's already a short list of some apps here that I maintain. It'd be neat if there was a way to input a URL to an app for it to be installed.
-
From what we have seen, when we mark something as unstable, people still expect support from it. When update fails, they want us to fix it despite it being unstable since they started using the app. One cannot make changes to app packages just because it's unstable. For example, if I publish odoo today as-is as unstable and without tests, is it OK if the next package breaks it? I think not, and you can see the statping, calibre-web, commento etc as examples where it just eats everyone's time.
One idea is that we have an ability to publish apps but they will not be listed in the app store. You have to explicitly put in like -
https://my.domain.com/#/appstore/com.newapp.cloudronapp. Apps will update as always. Is this something that is interesting? So, all one needs is the appstore id and from our side we can just push things without even really testing and not spending too much time. In my mind, since it's hidden, people probably think a bit more than unstable but I don't know
-
@girish I suppose unstable can be re-framed. I think setting expectations and hiding those apps initially would be good. For example:
- The app store has a setting where one has to go to enable the "unstable" section of the store.
- Enabling this has a warning modal asking people to accept all risks associated with installing these.
- Plaster notices everywhere letting people know that support requests for those apps will be ignored.
It is most definitely an issue of expectation management. Working in support, this is something I have to have front of mind always, as anything I do can set a precedent I may not want set. Therefore I have made a point of always being as clear as possible about what I will and will not do.
I guess my proposal of a fan-run unstable app store would step on your toes a bit but it was something I was going to make for myself to basically list apps that point to docker containers I build and then add the ability to use the API to install those apps to any instance.
-
From our point of view, we're more likely to be able to find time to contribute to fixing things if we've started trialing and expecting to use them in production.
It's not that we can't get apps packaged, it's just not our core business. Our business is implementing what's already available, to focus on business logic that pays for us and apps, platforms, servers, licences, and maybe something left for fun stuff.
I'm happy to share dozens of WP plugins because we had to develop them to get a job done and no-one would know they have any use until we make them exist.
The thing with app packaging, is they are all things that have a proven market and need, usually pretty closely correlated to their GitHub Stars count. So, we're not trying to prove anything to the world, just preserve time for the utility aspect that pays for other things we don't have any business in.
I guess I see it like a utility, I need electricity or WiFi, so does my neighbour, but it gets awkward if ask to borrow theirs.
I've forgotten where I'm going with this - but it feels counter-productive for me to do things that others can do faster at the expense of not doing the things I'm very fast at.
-
https://git.atrilahiji.dev/atrilahiji/cloudron-n8n-app
Works
Gotta use chrome/chromium based browser it seems. Firefox gets that connection lost issue.
I did only test with a webhook trigger going to an http request action (which is 99.9% of my use cases) but ymmv. I didn't test much else because I don't use a large number of the services it works with.
This is based largely on @jimcavoli 's package (I literally changed one thing in the manifest).Install by pulling the repo and doing the while cloudron build/install stuff. Ya know the drill now.
-
@atrilahiji can confirm I'm no longer seeing the 'Connection Lost' on Chrome! (Though still present on Firefox
). Either way, you're my hero! Will be playing around with this more over the next few days. -
@atrilahiji I came across a few things, wondering if you have some input on these:
- I'm not able to simply 'run' workflows. Trying to 'Execute' results in an endless spinning of a wheel. When trying to hit 'Stop' to stop the execution, this shows up in the logs:
Apr 27 13:51:02 Error: The execution id "6" could not be found. Apr 27 13:51:02 at /usr/local/node-12.16.3/lib/node_modules/n8n/dist/src/Server.js:1142:27 Apr 27 13:51:02 at runMicrotasks (<anonymous>) Apr 27 13:51:02 at processTicksAndRejections (internal/process/task_queues.js:97:5) Apr 27 13:51:02 at async /usr/local/node-12.16.3/lib/node_modules/n8n/dist/src/ResponseHelper.js:76:26- I can switch over a workflow to a webhook, and then properly trigger the workflow, which is awesome! However, to test this out, I'm doing a simple 'send email' using my Cloudron SMTP credentials. This node fails on the workflow with the following error:
ERROR: 140532647925632:error:1408F10B:SSL routines:ssl3_get_record:wrong version number:../deps/openssl/openssl/ssl/record/ssl3_record.c:332:
I know this is very much a work in progress, just wanted to record these in case anyone has some input.
-
@thetomester13 huh openssl errors??? O_O
Yeah not immediately sure why but I'll do some digging.
I'm also just going to do some cleaning up of the package
-
@atrilahiji upon further investigation , it seems as if it's just a problem with the Send Email node (or maybe my credentials? Though IMAP is confirmed working). I'm able to create other workflows with other triggers that run properly! Very awesome stuff!
-
@thetomester13 Yeah seems to be a weird upstream issue. I also just disabled SSL for SMTP and it worked. Still using port 587
-
@thetomester13 I also bumped both the N8N and Node versions
-
@thetomester13 said in Sponsoring: n8n:
@atrilahiji upon further investigation , it seems as if it's just a problem with the Send Email node (or maybe my credentials? Though IMAP is confirmed working). I'm able to create other workflows with other triggers that run properly! Very awesome stuff!
There seems to a bug with the cert renewal logic which I am just investigating. If you go to Services -> Mail -> Restart it copies over the fresh cert.
-
@atrilahiji @thetomester13 so, should i just go ahead and push that as unstable?
-
@atrilahiji said in Sponsoring: n8n:
@thetomester13 I also bumped both the N8N and Node versions
I'm still not able to get my SMTP Send Email node to work for some reason, even with the refresh of Mail certs and Ignoring SSL Errors. shrug not a huge deal for now, I'll play around with it some more coming up.
@girish said in Sponsoring: n8n:
@atrilahiji @thetomester13 so, should i just go ahead and push that as unstable?
I think that should be fine!
-
@girish It seems stable enough to be unstable. FYI the most up to date package is here https://git.atrilahiji.dev/atrilahiji/cloudron-n8n-app
