After recoverying a Backup file on new "On-premise" VM - Site no longer works

Product Support Self-hosting

AI-generated summary

A user attempted to recover a Discourse backup on a new on-premise Debian VM, but encountered a white login page and Redis connection issues. Initial troubleshooting included stopping Nginx, rebuilding the app, and restarting containers, but Redis was not found. The user discovered that Redis should be managed within the Discourse container, not the host system. After manually installing Redis and resolving certificate issues, the site became accessible. The user then faced a recurring white screen issue after another recovery attempt, which was resolved by transferring the backup file via WinSCP. The user credited the support team for guidance, particularly regarding potential plugin mismatches and CSP settings, and acknowledged the help in resolving the problem. ITAS-Isti documented the process and shared screenshots of errors and solutions throughout the discussion. Benjamin_D provided key troubleshooting steps and suggestions, including checking the app.yml and considering plugin compatibility.

1

Hi,

I habe recovered a Backup File from the cloud instance. After recovering it on my own debian server, the URL / Title of the page can be seen the login page is however white.

  • “Safe Mode” does not work.
  • Browser Cache is cleaned.
  • I have rebuilt the app, no difference.

Do you have an idea, what still could be done?

These errors have been reported according to the browser console:

report
report1657×835 390 KB

I have checked it in the production log as well, The connection to REDIS Servers seems not to work.

image
image1913×376 78.7 KB

I have deleted the container, rebooted the VM and installed Discourse once again. I have received this error:

Docker: driver failed programming external connectivity on endpoint app …

image
image941×226 132 KB

Thank you for your suggestions.

I.

2

you might have something listening on the 80 port (nginx ?)

3

I have stopped nginx and rebuilt the app.

According to production log REDIS doesnt work´:

image
image1555×306 108 KB

Do you have any idea what to do? I have already rebooted, rebuilt, container destroyed and it newly pulled etc. nothing has helped …

4

:thinking: sudo systemctl status redis ?

5

Redis is not found …

image

How can I start redis in the discourse container properly?

If I try to start it, it doesn´t work:

image

regards

I.

7

I have been actually able to install it first with a help of a tutorial. After I have recovered a backup from a discourse-hoster, I have received only white pages. Since that time I am not able to install it …

It can be that my reboots, rebuilds have crashed the vm itself.

9

How did you do that ?
./discourse-setup again ?

11

Debian 10.5 (Buster)

12

I have made it, many times, of course.

13

I think the problem is that redis works no longer and I do not now how to start it.
I have already pulled the discourse container, it has not helped however.

I think
I destroy the container,
delete the directory,
pull the container from git
and start from scratch again …

14

As you see, I have made it as simple as possible:

image
image970×400 80.2 KB

15

Indeed, I thought discourse-setup took care of the redis install :thinking: I guess you could try installing it “by hand” and follow what is done discourse_docker/image/base/install-redis at main · discourse/discourse_docker · GitHub

What is below your screenshot though ?
Any error after ./discourse-setup ?

16

The endresult of the installation was:

image
image1690×341 138 KB

I will try to install redis manually …

17

well redis seems ok here :thinking:

18

image

Discourse should work now! (it doesn´t)

Docker seems to be OK:

image
image572×776 64 KB

According “.launcher logs app” the issue must be with the certificate …:

image
image1321×535 170 KB

19

I’m sorry, my mistake for the systemctl check, I do have redis installed on the host but for another purpose, the redis you’re looking for is installed in the container.

21

Ok, one option would be to get a new Ubuntu VM (it costs however pretty much for our institute).
I think it is best solution just now, isn´t it?

22

that could come from your app.yml, what do you have there ?

## Uncomment these two lines if you wish to add Lets Encrypt (https)
#  - "templates/web.ssl.template.yml"
#  - "templates/web.letsencrypt.ssl.template.yml"

you might also have hit a letsencrypt’s renew limit…

26

no, I am “root” here. It is internal VM, no public access /up to now/

27

Thank you, it was the solution… :slight_smile:

Now I can access the GUI. I am curius, if the recovery of the backup crashes it again.

Are there any preparations, being necessary for an successful recovery from an other server?