Caprover... mysterious fails - Cloudron just works!
-
Thanks for the happy note!
-
Caprover is fine for testing various apps, experimenting and whatnot.
For mission critical operations, nothing comes close to Cloudron.I have a caprover too since it has Odoo. But I have to reboot it every other day, since it just stops working randomly without any error message.
I look to the day I can run Odoo on Cloudron.
-
@scooke my caprover generally is stable and works, but there are some 1-click apps which don't install well.
I added portainer to my caprover box which helps a bit to manage it.
I
% agree that Cloudron is rock-solid and my preferred deployment. -
Another similar effort is FLAP. Seeing as it's just one person, I think, running it, it's pretty impressive. There are still updates to its repository. He is open to direct feedback. Its GUI is not as polished as Cloudron, but I like it a bit more than the sparseness of Yunohost.
-
Gotta say, there's a lot more apps offered in Caprover I could do with for testing:
+1 for open-source ultimately attracting more development.
Cloudron's App Store development looks like it's gone stale, and I guess the community isn't big enough in comparison to pick up the slack.
What do we do?
Web Design https://www.evergreen.je
Development https://brandlight.org
Life https://marcusquinn.com -
@marcusquinn the CapRover 'catalogue' is interesting but beware .... many don't work.
They may have worked once, but a lot are not maintained, so unless you're happy with old versions, the catalogue shrinks.Support is not great - anyone can package anything, but caprover doesn't support anything. It's not as vibrant a community as Cloudron IMHO.
I like concept of Caprover and use a few apps from there, but I'm likely to shut my instance down.
-
@timconsidine Given their claim that it's using vanilla Docker, and all apps should continue working even if Caprover is removed, I expect the same issues would be present with just the Docker Compose scripts. We'll give it a try and see how it compares for time to package and maintain apps.
-
@marcusquinn said in Caprover... mysterious fails - Cloudron just works!:
Cloudron's App Store development looks like it's gone stale, and I guess the community isn't big enough in comparison to pick up the slack.
Maintenance is underrated
-
@girish And always appreciated!
I found some cool software for Mac recently called Shottr. One-man development, freeware, but with a link to BuyMeACoffee to allow for extra lubrication of the developer each time there's a nice update with new features & fixes.
Perhaps not likely to fund the sort of development needs of enterprise apps, but money does help in scaling teams and contributions, and currently I still don't see a way to help with funding over and above the basic Cloudron subscriptions. Enterprises will pay for apps and features, but no-one can do that gratuitously without having a way to do that, and an invoice to match that says what a payment was for.
Perhaps there's a way to do that with NodeBB: https://github.com/psychobunny/nodebb-plugin-paypal-subscriptions
Or you need to tap into the contribution features of GitHub, which seems to be better understood by developers, communities and contributors alike. Material for MKDocs sees to have created a sustainable and growing system for this. Perhaps some ideas here:
-
Just for reference, this is the update queue:
Each of the above app updates above breaks in some way (in fact, some of those updates are already weeks/months due). It's because all this is just super time consuming/exhausting and ultra technical. Most importantly, it's not at all fun and thus practically speaking is not solvable by buying someone a beer/coffee
In practically all cases, the issues are not Cloudron related but will affect any packaging system (compose, source installations etc). -
@girish Interesting insight. I continue to champion the cause and promote Cloudron to all developer contacts I know — in the hope that some pick it up and contribute to the community, and perhaps later their development experience to apps & platform.
Sounds like the parallel need is attracting more "ultra technical" wizards to assist, or somehow incentivising the app authors themselves to adopt the maintenance as part of their release testing. Tell us what you need - you have a community here for a reason, and there are many that want to help in whatever way they can.
Web Design https://www.evergreen.je
Development https://brandlight.org
Life https://marcusquinn.com -
@girish said in Caprover... mysterious fails - Cloudron just works!:
Just for reference, this is the update queue:
looking forward to Mailtrain 2 eventually hitting the app store one day!
-
@marcusquinn said in Caprover... mysterious fails - Cloudron just works!:
Sounds like the parallel need is attracting more "ultra technical" wizards to assist, or somehow incentivising the app authors themselves to adopt the maintenance as part of their release testing. Tell us what you need - you have a community here for a reason, and there are many that want to help in whatever way they can.
There is a general neglect of deployment and upgrades in the market which is far beyond cloudron. It's probably not apparent here (in the forum), but we regularly report bugs upstream where basic things like login doesn't work after a release.
Part of the issue is that in most projects the upstream deployment strategies are "automated" and nobody is really looking into what it built. There are no tests, I am yet to encounter a project which has proper UI and deployment tests. This is mostly why we are having to do this
I remember trying to upstream tests in our early days but the issue was that the tests are tied to Cloudron.There is also a lack of understanding of how deployments has to be done in a maintainable way. Many projects have plugins/extensions and I am yet to understand how people update things properly when they use plugins. The recent n8n/immich etc even start writing in code directories (probably bugs, but we have to investigate each thing and report upstream).
-
@girish said in Caprover... mysterious fails - Cloudron just works!:
There is also a lack of understanding of how deployments has to be done in a maintainable way.
Have you come across a good guide that one could share more easily with the project maintainers?
Many are asking for feedback and not getting the right kind, and they also don't know any better to go looking on their own.
-
@robi Closest I have come across is https://12factor.net/ which I think is from the Heroku folks. That document largely inspires how we do deployments as well.
-
@girish That's more for SaaS apps and not detailed enough for what the devs you're dealing with need.
So we need a deployment guide, perhaps specific to Github, that takes it a few steps further than the mere setup of the automation that makes it feel done.
The guide on tests and automation is another beast I think, needing a separate resource covering deployments and perhaps another for UIs as that's a whole other layer of QA after deployment.
Maintainability is yet another guide which gets specific to the project, yet is easier if the above two are already addressed sufficiently.
So that's at least 3 separate resources to share so far.
Here's what I found so far:
- Deployment Guide
- Tests and Automation
- Deployment maintainability could best be approached from your viewpoint which would reflect the needs of Cloudron, and hence suggest you write a few thoughts there and publish it so you can stop repeating yourself
-
I use both Caprover and Cloudron simultaneously, but honestly, Cloudron operates more stably. Caprover is quite difficult when installing many applications and allocating resources, and backup is not intuitive. All of that is available in Cloudron, the updates are timely and stability is high, support is quick even if using the free version. I just wish to develop quickly, install more applications and have the budget to upgrade to their paid package.
-
And a lot of the Caprover apps have problems installing/running when you use more recent versions of the core repos than the initial package was built for.
I've just killed my Caprover and will re-purpose the VPS.
Too little return for too few viably deployable apps and too much hassle.
