Quick CSS, Merged CSS & Object Caching issues

Good Morning,

once again a very nasty one :-) Happy for any input on how to deeper debug and solve this one.

The Setup:
We run the Enfold 4.8.6.5 on WordPress 5.8.1. Multisite. We have Object Caching up and running using Pantheons WP-Redis.
We have the issue, that the QuickCSS does not get correctly merged into the avia-merged-styles-…—….css file.

Our debugging shows the following facts:
– Saving QuickCSS correctly renders the content into the /wp-content/uploads/sites/…/dynamic_avia/enfold.css file
– This also triggers the generation of a new avia-merged-styles file
– Yet only the uniqueid() part changes, the file content hash remains the same. So does the content. The new QuickCSS does not get into the merged css
– Disabling the CSS merging outputs the corrent and up to date enfold.css with the updated QuickCSS
– Digging into the Object Cache we found a cache key prefix_12:options:aviaAsset_css_filecontent which does NOT get updated on saving the QuickCSS.

We believe that inside the asset-manager.class.php the function get_file_data might have some issues recognizing the changed enfold.css. Or to be more precise: To correctly handle updates of the option and a possible object cache. At least that’s the code part where the aviaAsset_css_filecontent option is conditionallywritten / updated.
After we delete the key from the object cache manually, the change is propagated correctly from the QuickCSS to the merged CSS file.
Anyway I am surprised, that the file key still remains the same although the content changed. Only the uniqueid changed. But I didn’t look into how you generate that key currently. I just believed it would be the hash of the file content which it obviously is not (anymore?).

Notable: Up until now (5-10 Minutes after deleting the object cache key) the cache key was not created anew. Might this have been some kind of a relict / leftover?

As you have a better understanding of this complex part of Enfold, we are very happy for any suggestions regarding the next debug steps.

Thanks and best regards,
Jan