What's coming in Cloudron 9.0 (was 8.0)

girish

@d19dotca don't have an ETA yet. We still have many apps to port over to the base image before we can make a release. I will update this thread end of the week when I will know better how much ground we have covered.

malvim

Sounds awesome! You guys rock.

I've had an interest in multi-host support forever, but haven't been following the discussions as of late. Curious about how billing and support will work.

Is this a single dashboard for multiple cloudrons, so that we'll pay for each one separately? Or is it a single cloudron with multi-host capabilities, so that we just throw a bunch of machines on a single cloudron install and it just does its thing?

Thanks!

girish

@malvim said in What's coming in Cloudron 7.0:

I've had an interest in multi-host support forever, but haven't been following the discussions as of late. Curious about how billing and support will work.
Is this a single dashboard for multiple cloudrons, so that we'll pay for each one separately? Or is it a single cloudron with multi-host capabilities, so that we just throw a bunch of machines on a single cloudron install and it just does its thing?

We want to implement the latter i.e it's basically a single cloudron with multi-host capabilities, where you can add nodes to a single Cloudron install. As for pricing/billing, there will be some per-node pricing component to our existing pricing. I don't have the exact details yet.

marcusquinn

@girish The most interesting thing about this is the way most hosts charge less per CPU/RAM for the smaller VPS tiers to capture people price-comparing.

It might be something that pays for itself, where growing single VPS CPU/RAM can be exponentially costly.

girish

Most of the apps are updated to the new base image, yay! That was a lot of work but atleast we can be confident that the apps work with the latest databases. So, we will push out a 6.2 release very soon with the following:

Move apps to new base image v3 - See https://forum.cloudron.io/topic/4366/docker-base-image-3-0
Database upgrades - MySQL 8, Postgres 12, Redis 5, Mongo DB 4.2.
Update node to 14.15.4
Update docker to 20.10.3
Make it possible to restore without updating DNS. This is useful if you want to test out cloudron backups.
Add a way to manage apps <-> groups better. Currently, it's hard to get a grip on what apps a user/group has access to easily (maybe add filters as well).
TLS addon (for DoT in AdguardHome)

girish

Just finished up on the "Dry run" feature for Cloudron restore. You can use this to test your backups, test migrate to another server to see how well the server performs (for example, when you switch VPS) etc.

The way it works is :

you select this "Dry run" checkbox. When enabled, it will restore as always except it won't update the DNS.
Before you click 'Restore', add an entry in /etc/hosts for my.domain.com to this new server's IP. You probably also want to add entries for your app domains if you want to test them.
Once restored, and you feel happy, you can "switch over" to this new server, using the new sync DNS feature. This is in the Domains view.

ruihildt

That's great!

Will you potentially lose some data during the backup/restore time that won't be carried over to the new one if you're running "dynamic" apps (like a forum)?

imc67

@ruihildt I was thinking the same! For example emails getting in to the old server will be lost I guess.

girish

@ruihildt yes, any data generated between the switch over is lost.

imc67

@girish wouldn't it be a good idea to have something like a "pause switch" all it does is (don't know technically):

"freeze" all the containers,
then make a backup,
migrate to new server,
do some testing and
switch DNS
finally "un-freeze / un-pause" all the containers on the new server (leaving the old one freezed)

Between 1 and 6 you can show the now already available and customizable custom_pages and incoming mail is already buffered at the sender.

d19dotca

@imc67 This is easy to do manually in the meantime, but yes I agree it'd be awesome to have this process sort of automated. For me it's mostly just email I focus on that's likely to change between backup and restore, so I do a full system backup and once it gets to the "box" part of the backup I disable the mail container so that it won't receive mail after the backup. Helps a lot. Of course doing this if you have clients relying heavily on email like I do then it means you have to do this in the middle of the night though lol. The ability to automate this process would be awesome.

girish

@imc67 Ah yes, we used to call this "retire" back in the managed hosting days We used to show a page that the server is migrating. Incidentally, I just removed the retire script last night! It's really just systemctl stop docker box but of course these days you will see a "app is down page" instead of a better "migration in progress" page.

I have created https://git.cloudron.io/cloudron/box/-/issues/774 but it won't make it to this release.

d19dotca

I installed and ran 6.2 this evening. So far it's looking great. Loving the changes and improvements. Especially happy with the 'missing backup' cleanup task So far so good! Thanks for all the hard work you guys!

One word of caution for anyone who hasn't upgraded to 6.2 (pre-release) yet... the downtime experienced during this upgrade was significantly higher than normal. Usually upgrades of Cloudron have practically no noticeable downtime or if it does it's only for a couple of minutes, but this time the upgrade took around 20-25 minutes before everything was running again which I'm sure (based on the logs) was because of the major version changes for MySQL, Postgres, Redis, etc. So just be prepared for a larger downtime during this upgrade.

Edit: I may have spoken too soon with regards to backups... running into two different issues with it now after upgrade to 6.2.

girish

I can confirm the bug. Those on OVH object storage for backups should not update to 6.2. Note that only the backup entries in the local database are removed, the backups itself are intact in OVH Storage. However, this means that Cloudron won't "clean up" the backups in remote storage properly since it has lost track of them.

Fix is coming in 6.2.1.

humptydumpty

@girish I closed the upgrade notification by mistake. What's the changelog for v6.2.1?

Edit: Nevermind! I found it in the "Event Log".

    "version": "6.2.1",
    "changelog": [
      "ovh: object storage URL has changed from s3 to storage subdomain",
      "ionos: add profit bricks object storage",
      "update node to 14.15.4",
      "update docker to 20.10.3",
      "new base image 3.0.0",
      "postgresql updated to 12.5",
      "redis updated to 5.0.7",
      "dovecot updated to 2.3.7",
      "proxyAuth: fix docker UA detection",
      "registry config: add UI to disable it",
      "update solr to 8.8.1",
      "firewall: fix issue where script errored when having more than 15 wl/bl ports",
      "If groups are used, do not allow app installation without choosing the access settings",
      "tls addon",
      "Do not overwrite existing DMARC record",
      "Sync dns records",
      "Dry run restore",
      "linode: show cloudron is installing when user SSHs",
      "mysql: disable bin logs",
      "Show cancel task button if task is still running after 2 minutes",
      "filemanager: fix various bugs involving file names with spaces",
      "Change Referrer-policy default to 'same-origin'",
      "rsync: preserve and restore symlinks",
      "Clean up backups function now removes missing backups",
      "Avoid updown notifications on full restore",
      "Add retries to downloader logic in installer"

? Offline

@p44 Multihost support sounds amazing. It'll make Cloudron unstoppable I think.

p44

@af

@A Former User said in What's coming in Cloudron 7.0:

@p44 Multihost support sounds amazing. It'll make Cloudron unstoppable I think.

Totally agree on this point!! I'm sure Cloudron team will do an amazing work!

girish

We started rolling out 6.2 over the last week, I think it will take until mid next week for the full roll out. As a heads up, there are two known regressions:

Renew all certs button does not work in the Domains view.
Linode Object Storage backups does not work . Trying to work out a fix upstream - https://github.com/aws/aws-sdk-js/issues/3404

humptydumpty

@girish I was about to post a "bug" about certs. One of my domains failed to renew 13 hours ago. I received an email from my CR about it.

{   "domain": "www.mydomain.com",   "errorMessage": "Unexpected status: invalid" }

I checked the logs and here's the timeline.
13 hours ago - certificate renewal failed
12 hours ago - certificate renewal succeeded
1 hour ago - certificate renewal succeeded (it ran it again?)

{
  "domain": "mydomain.com",
  "errorMessage": ""
}

Checking the cert on the site itself, shows that it did renew it successfully.

Edit: BTW, this is the same domain that I receive random messages saying that DKIM records aren't set properly but I know they are. The message comes back up a few days later. DNS manually controlled for this domain in CR.

girish

We will make another pit stop at Cloudron 6.3 before Cloudron 7 - https://forum.cloudron.io/topic/4723/what-s-coming-in-cloudron-6-3

Cloudron makes it easy to run web apps like WordPress, Nextcloud, GitLab on your server. Find out more or install now.

Cloudron Forum

What's coming in Cloudron 9.0 (was 8.0)