Issue Bug Under Review Quiet

#106298Malform URI in siteconfig takes down entire typo3 instance

Morten Haggren 03/04/25, 11:31 updated 10/10/26, 07:56

Description

Sumary

it is possible to save a site config with a ‘base’ / ‘Entry point’ that is malformed, doing so will crash the entire typo3 instance to a point where recovery requires direct file access outside of typo3.

Steps to reproduce

  1. edit or create a new site config, enter a malform uri ( “//https:domain.tld/” ) in “Entry point”
  2. Save the config

The typo3 instance is no longer usable ( frontend(s), backend or install tool ) – the error is ‘InvalidArgumentException: The parsedUri “//https:domain.tld/” appears to be malformed’

This error originally occurred in a LTS 10 instance, but can be reproduced in LTS 13 as well.

NB. In testing the installation doesn’t seem crash if there is a valid ‘Variants for the Entry Point’ that would be active

Desired outcome

The user should either be prevented from saving a site config with an invalid/malform uri or the site config should be ignored entirely if it cannot be validated, this way at least the user has a chance to correct the issue in typo3.

Other notes

The user swears that they copied the url from their browser and pasted it in the field, so it’s possible that it somehow got mangled during the save. ( eg. if pasted after/inside an existing URI )

Reviews 1

History 812

  1. Georg Ringer commented and changed status Under Review Closed <span data-processor="relativetime" data-value="2025-03-17T21:21:56+00:00" > 03/17/25, 21:21 </span>

    closing as duplicate of #106298 which got a potential patch attached

  2. Morten Haggren commented <span data-processor="relativetime" data-value="2025-03-18T07:27:30+00:00" > 03/18/25, 07:27 </span>

    Georg Ringer wrote in #note-4:
    > closing as duplicate of #106298 which got a potential patch attached

    I would like to request that this be reopened, as closing the issue as a duplicate of it self seems like a very strange move

  3. Garvin Hicking commented <span data-processor="relativetime" data-value="2025-03-18T07:30:38+00:00" > 03/18/25, 07:30 </span>

    Sorry for that, this was not intentional :-)

  4. Florian Seirer commented <span data-processor="relativetime" data-value="2025-09-22T10:49:07+00:00" > 09/22/25, 10:49 </span>

    Same here on TYPO3 12 and 13, but I don’t even have to save the config for this to happen. I just clear the caches, and then sometimes (not always!) the frontend and backend is dead until I delete /var/cache/* manually.

    We use “https://%env(SERVER_NAME)%/” as base in config.yaml, if that matters.

  5. Markus Klein commented <span data-processor="relativetime" data-value="2025-09-22T11:05:30+00:00" > 09/22/25, 11:05 </span>

    We use "https://%env(SERVER_NAME)%/" as base in config.yaml, if that matters.

    I can only guess, but we encountered that before:
    You probably have a cron-job running that is executed before some web request hits the system the first time after cache flush.

    Via CLI there is no SERVER_NAME hence you end up with a broken site config. (check cache folder there is probably a base `htts:///` in there.

  6. Florian Seirer commented <span data-processor="relativetime" data-value="2025-09-23T06:26:42+00:00" > 09/23/25, 06:26 </span>

    You might be right. Thanks for the hint, Markus!