Cloudron completely broken after update/reboot

[JohnBee]

Came into the Cloudron interface and found a reboot required message

Did the reboot, and then found ipv6 broken - I was able to address this using the following approach;

sudo sysctl -w net.ipv6.conf.ens18.disable_ipv6=1
sudo sysctl -w net.ipv6.conf.tailscale0.disable_ipv6=1
sudo systemctl restart box

That said, and after a reboot, I then found the Mailserver service was completely down both internally and externally - a closer examination revealed a mail container corruption overlay filesystem reference, preventing the service from starting - ie, mail configuration directory at /home/yellowtent/platformdata/mail/ was missing, preventing automatic recreation by the box service

Would also add that standard recovery attempts(sudo systemctl restart box), was not successful, in-that the mail container would be automatically recreated

A closer look revealed the configuration was missing;
sudo ls -la /home/yellowtent/platformdata/mail/

no such file or directory

And that a forced rebuild was needed;
sudo rm /home/yellowtent/platformdata/INFRA_VERSION

BTW, this same thing happened in previous months with Cloudron, following an automated update - granted, snapshots can(and should), be made priior-to any type of update/OS or otherwise, though I can't help but feel that issues like this should not occur so frequently, even on a personal use level

• hope this helps someone out there

[james]

Hello @JohnBee
This is rather odd that it happened twice already.
What is your provider for hosting the server and did you do any customization underneath Cloudron?

[JohnBee]

The first time this took place(update, notification/broken Cloudron services), was on a VPS hosted by SSDnodes - thus constituting my first experience with the infamous IPv6 bug, and as a result of a reboot request(notification), following a Cloudron update.

In this particular case, and though I did confirm the IPv6 issue, I chose not to repair, in-favor of rolling-back(snapshot), for the sake of brevity.

Fast-forward to this round - and on a rather fresh Cloudron installation, on Proxmox VM w/different provider, I'd add that both instances would constituted what I'd call vanilla, in-that they were both simple mail servers, with a single SOGo application for webmail.

Beyond that, there were no modifications made on either instances beyond the addition of nftables.

NB, the first instance was on a network with IPv6 support, whereas the second(more recent), is on an IPv4 only network.

PS, if you would like additional details on this particular incident, please don't hesitate to ask as the details remain fresh on my purview, and moreso, that I would really like to help in improving the Cloudron experience for others

[james]

The more details the better.
This is very uncommon.