ragnarkurm → created an issue.
ragnarkurm → created an issue.
Fixed file paths.
I ended up with a bit simpler patch but it goes without tests.
The patch works but has unintended side-effects.
Namely, the new parameters `message_orig` and `message_args` leak into the logs.
Here is how the log looks now:
{
"@timestamp": "2024-06-01T18:57:01.471Z",
"@version": 1,
"message": "event: unpublished asdf",
"message_orig": "@type: @action @title",
"message_args": {
"@type": "event",
"@action": "unpublished",
"@title": "asdf"
},
"message_id": "4a5b318cb696665b6efd72f120.96887480",
"site_id": "4a5b318cb696",
"canonical": "...",
"method": "GET",
"tags": null,
"type": "drupal",
"subtype": "publishcontent",
"severity": "Notice",
"request_uri": "...",
"referer": "...",
"uid": 53,
"username": null,
"client_ip": "...",
"link": null,
"code": 0,
"trunc": ""
}
Simple manual testing seems to indicate that minifyhtml appears to break Big Pipe too. Some Big Pipe items are not loading. Nothing in Network not Console tab about this. The HTML seems to be broken. After disabling the minifyhtml along with the patch, the ending of the page is now good.
minifyhtml installed and patched:
minifyhtml uninstalled:
The patch works in a sense that pages are loading now.
But in Nginx logs there are now following errors:
upstream sent more data than specified in "Content-Length" header while reading upstream
It seems not to happen on every page. I didn't do a thorough testing, but it seems to happen on node pages and not in admin pages.
ragnarkurm → created an issue.
ragnarkurm → created an issue.
ragnarkurm → created an issue.
As Google search is leading here, more people might be looking for help regarding "too many realms/grants", I added the info here. Encountered the issue with the permissions_by_term module, but the problem is more generic. Created also a core patch to alleviate the problem. The core patch is in the module issue queue since I understand that core maintainers are not willing to optimize NodeGrantDatabaseStorage::access(), and pointing toward using more specific hooks.
https://www.drupal.org/project/permissions_by_term/issues/3380033 📌 Performance degradation Postponed: needs info
The patch in the issue takes care of monster select queries to the node_access table per node, but not for views. It is still a noticeable improvement in terms of speed.
ragnarkurm → created an issue.
ragnarkurm → created an issue.
ragnarkurm → created an issue.
@EricVL - your solution worked. Thank you!
We are experiencing this problem on NFS based filesystem.
What seems to be happening is this:
- Drupal creates a file
- NFS starts to sync file creation operation to the NFS server
- Drupal tries to touch the file, but NFS thinks the file is not there (yet)
- Drupal issues error/warning
- NFS completes file creation operation (say in 1 sec), and the file is now seen on the NFS client system
- Note: the file sometimes has a random
mtime, usually far in the past