What's the largest Cloudron you have seen?
-
Fun question for a sunny day... how large is your Cloudron or one that you have come across?
For context, I am trying to figure out whether I want to scale vertically or keep scaling horizontally with my Cloudron. I have tried running two and would prefer the reduced maintenance of a single one... because... er- efficiency
Issues I can see are much longer backups and restores, if they were needed. Although, I use Cloudron backups mostly for migrations (very occasionally) and to manage individual apps. I use VM snapshots for disaster recovery (DR) purposes as they are much quicker.
What's your experience? How big would you go?
Let me know
-
We started with a Cloudron instance and used this instance for everything. Mail, web, apps, VPN ... Every time the instance needed a restart, we all crossed our fingers that it would be available again soon.
That was the moment when we decided to split the services into different instances. We currently use about 5 instances that are connected to a central user administration as βIAMβ (also a Cloudron instance). Incidentally, it is also better for Cloudron as a company because you pay for more than one license. That's what the (new) plan Business is for. https://www.cloudron.io/pricing.html
With up to 50 users, this works for us. Some of our customers have 400+ users. From this use case we learned that the Cloudron user interface doesn't have nice features for such large numbers of users, mail addresses and group filters. It works, but it's not that nice to use.
-
Mostly standard VPS at hetzner & co. An instance as a playground on a https://www.zimaboard.com/ In the early days, we used one instance as a backup server. That was bare metal with raid. https://forum.cloudron.io/topic/3508/show-me-your-dashboard?_=1745944465785
-
We started with a Cloudron instance and used this instance for everything. Mail, web, apps, VPN ... Every time the instance needed a restart, we all crossed our fingers that it would be available again soon.
That was the moment when we decided to split the services into different instances. We currently use about 5 instances that are connected to a central user administration as βIAMβ (also a Cloudron instance). Incidentally, it is also better for Cloudron as a company because you pay for more than one license. That's what the (new) plan Business is for. https://www.cloudron.io/pricing.html
With up to 50 users, this works for us. Some of our customers have 400+ users. From this use case we learned that the Cloudron user interface doesn't have nice features for such large numbers of users, mail addresses and group filters. It works, but it's not that nice to use.
@luckow said in What's the largest Cloudron you have seen?:
From this use case we learned that the Cloudron user interface doesn't have nice features for such large numbers of users, mail addresses and group filters. It works, but it's not that nice to use.
Yeah I don't even have a large number but clicking through pages of users (or domains) is annoying. Even though search also exists.
Hopefully @staff will address such issues in the current redesign.
As a bare minimum I want to be able to choose to display many more rows per page (e.g 25, 50, 100, 500 ie up to and including all users, or all domains etc)
-
@3246 We are using a company called SSDNodes. Our largest Cloudron is a 12CPU, 48GB, ~1TB SSD. We are hosting large mailboxes in addition to apps and wanted lots of extra space for growth. SSDNodes is U.S. based but have datacenters everywhere (https://www.ssdnodes.com/data-center). We are 5 months into a multi-year contract and zero issues:) I would only place "long-term" loads with SSDNodes as they require at least a 1-year commitment. The longer you contract, the more you save.
Here is a referral link if someone is interested: https://www.ssdnodes.com/manage/aff.php?aff=2481®ister=true (FYI, I will use any referral fees received to benefit the Cloudron community rather than me personally.)
-
We have removed the pagination in the new UI and will fetch more on scroll instead. Pagination seems a bit oldschool these days
-
Thanks for sharing details about your Cloudrons.
How do you handle recovery from a borked server?
I find that using the build-in backup not very good for speedy restores of an entire server (your milage may vary depending on sizes). That's why I use them for tactical restores or cloning of apps only, unless I have to move a server between providers (which I try to avoid). A local VM snapshot appears to be the fastest in my experience.
My RTO (Recovery Time Objective) is to have the server backup and serving within one hour. My most common scenario is likely to be an update or change of system configuration that results in loss of service. Considering that IP's don't change, this seems to work OK.
I'd love to hear your experiences
-
Thanks for sharing details about your Cloudrons.
How do you handle recovery from a borked server?
I find that using the build-in backup not very good for speedy restores of an entire server (your milage may vary depending on sizes). That's why I use them for tactical restores or cloning of apps only, unless I have to move a server between providers (which I try to avoid). A local VM snapshot appears to be the fastest in my experience.
My RTO (Recovery Time Objective) is to have the server backup and serving within one hour. My most common scenario is likely to be an update or change of system configuration that results in loss of service. Considering that IP's don't change, this seems to work OK.
I'd love to hear your experiences
@3246 said in What's the largest Cloudron you have seen?:
My RTO (Recovery Time Objective) is to have the server backup and serving within one hour
Yes depending on size and type of backup provider used.
Keep in mind, Cloudron will have multi backup destination soon (tm).I know of a setup (that I helped to set up) that uses a RAID 0 PCI 5.0 NVMe drives with Cloudron ext4 Backup provider with superfast backup and restore times.
Since, it is a local disk/raid directly connected to the system.Since an external backup is still wanted by him, this setup uses a custom script that syncs that RAID to an external provider. (note: not needed in the future).
I've seen this setup restore a 500GB App within ~60 seconds.
And the whole system less than 10 Minutes.You could go even faster with a RAMDISK (volatile) to RAID0 (local) to external backup chain.
But since the RAMDISK is volatile, on boot the data from the RAID0 would need to be re-synced to the RAMDISK.
Otherwise the Cloudron App or full restore would not find its data. -
Wow, @BrutalBirdie, that's some speed!
I am using VM's (currently Hetzner) and backup to StorageBox as well as VM snapshots. I had a bare metal box at Hetzner before, albeit with spinning disks of rust instead of the super fast NVMEs you are running. It took more like 6 hours to restore when I moved (on a weekend).
I look forward to having multi-location backups and used a similar means to achieve redundancy. I backedup to a local SAS HDD (ext4) and then ran a cron job to shovel a copy to StorageBox.
I wish we could have Borg backups (with multiple repos/locations)
@girish
-
My Cloudron details:
Hosting Provider: Netcup
Hosting Product: VPS 8000 G11 iv 12M MNZ (16 core, 64gb ram, 2048 SSD)
Total Apps: 158
Backup provider: Backblaze B2I and my company are all in on Cloudron.
I use it for my business and for my family office. In the process of packaging up a number of applications for Cloudron that a fully self-hosted business would need.
-
My Cloudron details:
Hosting Provider: Netcup
Hosting Product: VPS 8000 G11 iv 12M MNZ (16 core, 64gb ram, 2048 SSD)
Total Apps: 158
Backup provider: Backblaze B2I and my company are all in on Cloudron.
I use it for my business and for my family office. In the process of packaging up a number of applications for Cloudron that a fully self-hosted business would need.
@charlesnw said in What's the largest Cloudron you have seen?:
In the process of packaging up a number of applications for Cloudron that a fully self-hosted business would need