Roundcube Community Forum

 

Message editor issue

Started by core_dumped1957, August 15, 2025, 01:52:45 PM

Previous topic - Next topic

core_dumped1957

I have Roundcube Webmail 1.3.17 combined with Dovecot 2.2.36.4-1.
It has been working for years with no issues.
Yesterday, one user reported the editor window to write a new email would not come up and
eventually showed a "Gateway time out" message. The same with "Reply".
This happens in just one account. All the others work as usual, with no issues.

Things that have been tried, with no results:

1. Recreating the .Drafts folder
2. Emptying mailbox and using the account with zero messages.
3. Deleting the user from the mysql (Mariadb) database.

Monitoring the logs while trying to write an email produce no error message.
This is a strange problem because, as I said, Roundcubemail is an application I simply
never have to worry about.

Thanks for any suggestions,

SKaero

Where did you look for logs? There maybe something in <RC root>/logs/errors or since its a gateway timeout it maybe a problem with your web server so check those logs as well.

core_dumped1957

I did look for the logs at <RC root>/logs/errors. I have reproduced the error monitoring this file and no error is logged.

You are right, it could be the web server but, then it would happen to any account wouldn't it? This happens only with one account, across different browsers based in different pc's with Windows, Linux and MacOS.

I would think that something happens if there is one file, one email that has some kind of problem. But as I said, It happens with an empty mailbox. It only needs to be one specific account.

SKaero

We have to find the error to find the root cause, what is the web server setup on the server?

core_dumped1957

Mmm, the server is Nginx and I have finally found a log that sends an error when the RCM editor is invoked. It is in Nginx and it reports that:

2025/08/19 11:17:49 [warn] 18177#0: *108064 an upstream response is buffered to a temporary file
/var/lib/nginx/tmp/fastcgi/9/86/0000000869 while reading upstream, client: 10.21.48.88,
server: www.myserver.com, request:
 "GET /roundcubemail/?_task=mail&_file=rcmfile2121755623732025887000&_id=110322821068a4b12832b0f&
 _action=get&_extwin=1&_mimewarning=1&_embed=1 HTTP/1.1", upstream: "fastcgi://127.0.0.1:8011",
 host: "www.myserver.com", referrer: "https://www.myserver.com/roundcubemail/?_task=mail&;
 _preload=1&_file=rcmfile2121755623732025887000&_id=110322821068a4b12832b0f&_action=get&_extwin=1"

However, 'client_max_bdy_size' is set at 128M in Nginx configuration file. I would say that is quite large.

Thanks for your response.


SKaero

Since your using Nginx and its complaining about an upstream issue I assume your using PHP-FPM. Have you checked those logs?