After activating a plugin in a local development environment, and clearing the cache, I get a blank white screen with no error messages at localhost/admin, and a green status „200“ in Symfony profiler in the storefront.
I also ran bin/build-admin.sh, bin/console cache:clear, and cleared my browser cookies and logged in again into the administration. Same situation: blank white page, no errors, green profiler status.
There are related posts and reports of similar situations in the forum but they’re either about bugs that should have been fixed or about shop (core) updates, recommending to deactivate plugins, delete cookies, clear the cache etc.
After deactivating the plugin, admin is back again, so there is probably an error in the plugin, but neither the Shopware core, nor the activation scripts, nor the shopware-cli extension validate, nor the Symfony Profiler show any errors. The logfile /var/www/html/var/log/dev.log is full of the same deprecation warnings that Shopware 6.5 logs when everything works, but not helpful new messages about the critical plugin either. The only request.ERROR log entry is a MessageQueueException that another worker is already running for receiver.
There is one error in the browser’s network log:
http://localhost/api/_info/entity-schema.json returns 500 Internal Server Error.
How can a plugin kill the whole backend without any visible error message? What else can be done to find out what’s wrong?
What else can be done to find out what’s wrong: PhpStorm, with Symfony and Shopware support, shows warnings in the „Problems“ tab. I had an undefined class and an unused element after renaming my plugin’s extension class.
Still not sure why this
- does not trigger any warnings or errors in Shopware
- can break the backend without any error or fallback
- while the storefront seems completely unaffected,
but it’s the usual solution: inspect your code and make sure that all class names, files names, and service definitions make a perfect match, and take care about cases and capitalization etc. - and set the default log level to WARN, otherwise each request will write hundreds of lines deprecation notes to dev.log.
When the dev logs are silent, no more „problems“ in PhpStorm, verified that the admin works without the plugin, but the white screen comes back after activating the plugin, what else to do, as there is no Symfony profiler in the backend?
I kind of worked around the symptomatic problem. I reviewed my current code, still found no errors, but found out that I had conceived an overengineered entity structure that I found hard to maintain even in its initial stage. After simplifying my code and getting rid of redundant boilerplate code to define the same data schema in at least three different files, I refactored my code using custom properties.
I can’t reproduce the error anymore, but I can’t mark this as the solution either.
Hi!
Does the plugin extend or create an own admin module/component?
I faced similar problems in the past and the problem was about errors in the twig template of a page/component… unfortunately it does not show you the errors anywhere, it’s just a blank page then.
1 „Gefällt mir“
No, it’s a frontend plugin. It`s working now and has been tested and released in the Shopware store as Cost Transparency). As I said, I had not been able to reproduce the error, which is kind of worse than if I had, because I still don’t understand why it happened in the first place.
The only backend related configuration is done descriptively using configuration cards in config.xml:
it could have happened through a wrong configuration in for the components you are using in the config.xml as they will be compiled/used as VueJS components in the end.
But good that you don’t have the problem anymore