I recently installed NC 23 Hub II on a V-Server with 4 vCore and 8 GB RAM at the hosting service Strato.de. I installed the Collabora app and the CODE server. While loading an *.odt document I get the error message: âDocument loading failed. Try again later.â
RAM is enough but Iâm afraid that the server (virtual server) is not suffiend enough. I tried the Collabora demo server and that works fine. Back to CODE I get the same error message.
Sadly, my rented server is to small for that. But I donât use LibreOffice often, so I deleted it from NC. But that could possibly be the issue on the Raspberry Pies. Perhaps in some time I will use a bigger server (My whole family use it) and maybe CODE will works then.
Hi,
Iâm expiring the same issue - every try to open a file with the built-in CODE server ends up in error âFailed to load Collabora Online - please try again laterâ
Looking into the error console of the browser i see 502 Proxy Errors and therefor I looked directly onto the server. There I tried to run the Collabora_Online.AppImage as user www-data and Iâm getting a âpermission deniedâ error.
Looking at the file permissions i see: -rw-r--r-- and this seems pretty strange to me⌠Modifying the permissions lets me start the AppImage but then I get an other error (that maybe is connected to some other work happened lately on my server and I think this isnât happening to everyone).
Could someone maybe confirm that the permissions have changed with the last update of CODE to Version 21.11.103?
Hi, I recently open a threat with this problem, but I am not runing on docker or snap. I see that my permissions of my CODE built-in and collabor app are www-data:www-data and â-rwx r-x r-xâ at the root folders. The files inside are â-rw- r-- rââ with same user and group
I just solved my problem adding the line âallow_local_remote_serversâ => true,
to my config.php, restart apache (systemctl restart apache2) and now it is working!!!
Thanks a lot for your reply - unfortunately this didnât changed anything in my situation .
As before the last CODE Update everything has worked without problems i have a suspect that this could have generated the problemâŚ
The missing permission to execute the AppImage from shell is also strange - before the update this seemed to work. Maybe this has been changed intentionallyâŚ
Iâve found the solution for my problem: the process of coolwsd blocked the port on the server⌠therefore after the update the newly started AppImage couldnât connect to the portâŚ
Iâm sure theyâve created a frustrating confusing disaster for many people because who thinks to look at a working reverse proxy configuration? I donât know why they didnât just do a symlink in the container instead of wasting hours of peopleâs time trying to hunt down the problem but oh well.
Yeah, I first noticed the issue with the beta versions, but thought, eh maybe itâs an issue they will fix. Then they rolled out the stable and it was still happening, and I was thinking⌠crapâŚ
I happened to be digging in the docker container, cause I had scripts that deployed the container with additional fonts. That portion suddenly wouldnât work, which I found was because of the folder change. Then I was eventually able to put two and two together. But still, that was an hour and half Iâd like backâŚ
oh trust me im okay with wasting my time reading the official docs but the thing is, official docs are old even (uses the old lool) This is why it was super confusing like i followed the docs and yet canât do it
Slightly off topic - but docker collabora version 21.11.1.4.1
confirmed working. Iâm aware a lot of the other posts above were in relation to non-docker version, however sometimes its good to find information in general on what versions work, and what do not.
I think most know newer versions of docker work fine. The issue was the surprise, not really announced, extra reconfiguration of the reverse proxy server for those upgrading from versions earlier than 21
@jshpettus â Hey thanks for that information. Really good. Mine was running fine but after reading your response â I thought to myself â what the heck?? My configuration should be blowing up just like everybody elses. But it wasnât. So I had to investigate.
So my setup is a traefik reverse proxy (communicates via https with nextcloud) that proxies over http to the collabora docker container listening on port 9980.
I looked at the documentation and all I saw mentioned was either Apache or Nginx for the reverse proxy â nothing about traefik. I see where the developers switched the loolwsd directory to coolwsd (which honestly imo very dick move but whatever). Traefik since it was tied to exact directly structure that was modified continued to reverse proxy as normal. Weird. I guess truly dumb look on this one.
For the record for future users. This error can also be a result of security changes they made I think a year ago.
With newer versions of the container they have gotten rid of the old âdomainâ flag and replaced it with the optional aliasgroup# which lets you specify even multiple domains if needed.
Now the default, with no flag set, is collabora server will bind to the first connection of the MAC address of the Nextcloud server that connects to it. It will then not allow anything else to connect (and you will get that error). Thatâs fine for most, but depending on your set up, you may need to specify that aliasgroup1 flag so it can go by domain instead.