@grimreaper said in Mattermost vs Matrix/Element:

@jdaviescoates

You need to get updated, the community edition does has the ability to set permissions on mattermost.

screenshot-nimbusweb.me-2023.08.30-05_51_46.png

Good to know, thanks for the update.

These guys are good for that type of service https://account.dyn.com/

@jdaviescoates said in Cloudron update to 7.5.1 - Success!:

@mhgcic said in Cloudron update to 7.5.1 - Success!:

I could increase the RAM to see if that helps.

Regardless if it is what's causing the issues you're currently facing, I'd definitely get more RAM if I were you.

Each instance of Mastodon needs at least 1.5GB and each instance of Discourse at least 2GB. Already that is at least 10.5GB RAM you need before you've included all your other apps plus Cloudron and Ubuntu themselves.

And these are bare minimums. If you have users actually actively using these apps they'll need more.

We know this and monitor daily, and we have software that lets us know that RAM needs to be increased whether it is on a server level or on an app level.

We upgrade as and when needed, We also use throttling so if extra RAM or CPU is needed it is given until we decide on its upgrade needs.

After testing it had nothing to do with CPU or RAM it seems to be networking or a service possibly.

@girish Sure once I have some time I will do it. Thanks. 🙂

🤷 not sure

@David-0 said in Memory usage:

do the apps need this much memory all the time

Yes, at least.

I find that in practice in use they nearly always use more.

It's probably overkill, but I tend to default to manually setting the RAM to double the minimum.

@David-0 said in Memory usage:

Do I need to upgrade?

More RAM is always good.

But you don't really need more until you start regularly getting apps crashing because they ran out of memory.

@nichu42 that you can safely upgrade to

Thank you,
I got this solution

Cloudron is the ultimate flywheel business. More users -> more community -> more insight -> more relevance -> more traffic -> more users -> more investment -> more apps -> more relevance -> more community -> more users 🔁

The whole reason I invest in this forum, and the crafting of my posts with SEO considerations, is the ROI I get from its growth in capabilities and like-minds to compare notes.

It's like having the best sys admin team to work with, everyone's here for the results, the better results everyone gets, the more come here to help with more positive results.

All the wins from one very reasonable subscription / membership. Long may it grow, and become an increasing part of a healthier online future.

This topic was discussed few times already for example at https://forum.cloudron.io/topic/7231/cloudflare-tunnel

This week the forum was way more reliable for me except one or two cases where the forum software told me I have no internet connection. So I wanted to report this issue as resolved but then I faced just right now this (502 Bad Gateway errors; refreshing several times helped):

grafik.png

Interestingly, I also face at the same time issues with Microsoft Office 365 services. Looks like they share something in common.

Don't have experience with BitNinja but Cloudflare is used quite commonly for this.

This is correct, only the Cloudron version is relevant in this case

@CBCUN It's in our TODO list. I don't have an ETA yet.

@girish Oke feel stupid. That worked, thank you.

I’m using Tailscale IP, so all traffic is going via the tunnel. My assumption is that it doesn’t matter if it’s not encrypted as long as I use the Tailscale tunnel? 😶‍🌫️

In case, anyone is using borgbase, they seem to be having a big outage. Possibly down for a week already.

https://status.borgbase.com/status/default

I found out from https://news.ycombinator.com/item?id=37115540
image.png

robi said in [Guide] Auto Deploy of Hugo via Gitea + Drone CI:

how this could all run from the same cloudron

It already could. Drone's runner (would be the same with Gitea or Woodpecker) is a simple go binary that just needs access to a local docker socket. It's just me running jobs on my installation, so there's no risk of exposing data to others by accessing the Docker socket.

And if you're serious about CI, you'll want optimised machines anyway (building in a ramdisk ftw), not running your CI jobs where their resources are competing with Wordpress & co. Being a simple binary that doesn't hold any local data also means you don't need to back up your runners, and since they access a central node, they don't even need internet access.

If you want to save costs, you can even let your developers run the runner on their local machines.

You could also use rclone together with mergerfs to create some kind of layer.
like
/mnt/remote
/mnt/local
/mnt/mergerfs

You then write stuff locally, and it'll upload stuff in the background to your remote storage.

You can look for inspiration @ https://github.com/l3uddz/cloudplow

Some tar - streaming would help with all the small files.