Page MenuHomePhabricator

What happened to core.pkg.css?
Closed, ResolvedPublic


My browser can't load /res/phabricator/59ea1706/core.pkg.css. My phab instance is super ugly:

Screenshot_(11).png (942×889 px, 55 KB)

commit 3a81f8c68d2983231c978171135145a2c57a7e5f
Author: Chad Little <>
Date: Sun May 18 16:10:54 2014 -0700

commit 32f1dad1f4e5b41671ab0e5ec2ca51544b0546b9
Author: Joshua Spence <>
Date: Sun May 18 07:21:17 2014 -0700

Event Timeline

igorgatis raised the priority of this task from to Needs Triage.
igorgatis updated the task description. (Show Details)
igorgatis added a project: Phabricator.
igorgatis added a subscriber: igorgatis.

Side question: is there a concept of Last Known Good commit? A point which everybody can sync to which is known to be stable? I pulled a few minutes ago and just hit this. Now I'm stuck.

Is there anything in your error log? I just pushed this instance and it seems to be working fine (at the same commits).

Where should I look at? Sorry, still learning.

Sounds like it is generated, right? If so, is there a way to force regenerating it?

It's generated dynamically at runtime: the request generates it.

The error log is usually in /var/log/apache/error.log or /var/log/http/error.log or similar.

What response do you get when you access the page itself?

I'm using nginx (fwiw, apache was hosing my machine). Error file is empty. /var/log/nginx/access.log has a bunch of regular lines plus: - - [19/May/2014:00:59:40 +0000] "GET /res/phabricator/59ea1706/core.pkg.css HTTP/1.1" 404 2862 "-" "Mozilla/5.0 (Windows NT 6.3; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/34.0.1847.131 Safari/537.36"

Accessing it directly gives me 404.

git reset --hard 404cc16c887fca060b5c39ab914fd261d42e00d6

fixed it. I wonder whether future updates will break it again... is there any other log I could look at?

Does the next patch break it?

git reset --hard 1ec21dc81de596150dcf40218154b7e16a468d1d
igorgatis claimed this task.

Not really. :p

I binary searched over the last commits trying to repro and I reached HEAD without breaking it again.

That means I is solved for my instance, which is great! Thanks a lot!

I guess if someone hits it too, he/she can do the same to get it fixed.

Let me take that back. I guess I as fooled by the browser cache. I disabled the cache and binary searched again. It seems rP3a81f8c68d2983231c978171135145a2c57a7e5f is the culprit.

I updated my update script (downloaded from and modified to use nginx) and it looks like this:

cd $ROOT/phabricator
git pull
#git reset --hard 404cc16c887f
#git reset --hard 28c198d9026e
#git reset --hard d744d5d85995
#git reset --hard 7a9d5f8f2de5
git reset --hard 3a81f8c68d29

With rP3a81f8c68d29, I have a repro. When I revert to rP7a9d5f8f2de5, the problem is gone. Perhaps he change in map.php is related?

Are you still having issues at HEAD? We haven't received any other reports on this, but would be curious as to the cause of course.

I updated to HEAD successfully. I believe it may have been caused by a missing php restart.

bogdaniy added a subscriber: bogdaniy.

I have the same problem. My browser can't load core.pkg.css and core.pkg.js files, it is trying to load it for 30 seconds, and stops loading.
In my acces log file i have just one line about this files - - [24/Mar/2015:20:45:21 +0200] "GET /res/phabricator/cbdbd552/core.pkg.js HTTP/1.1" 200 24320 "-" "Mozilla/5.0 (Linux; Android 4.2.2; 2013023 Build/HM2013023) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.96 Mobile Safari/537.36" "-"
I can't reset project to old versions, because my db version doesn't support it. All my team members have this problem.

This isn't a problem with Phabricator. Usually, fully updating and following the update guide will resolve the problem. Specifically, make sure to stop and restart apache / php-fpm.

@chad I was thinking about that too, but after I updated phabricator this problem left unsolved.

This doesn't appear to be an issue with Phabricator, that is, it doesn't reproduce for myself locally or on this install. In most (all?) cases this is due to issues outside of Phabricator, such as proxies, apache configs, improper updates, etc. So I don't think we're going to be able to give you much in the way of useful advice. I'd start by doing a fresh update (and restart) and check error logs (apache, php).

I use nginx, and nginx error.log is empty. I have php 5.6.6 version, and php error log is empty too. I doesn't change any config files and I doesn't update my server software recently.
I can't understand why only pkg file doesn't loading, for example - 200OK - very long loading, and finally ERR_CONTENT_LENGTH_MISMATCH

@epriestley do you have any thoughts? I don't know nginx too well.

We use nginx on this server and the same resource loads properly:

I have no clue what's wrong.

@epriestley This is my nginx phabricator config
Is there any way to regenerate pkg files, or view generator logs?

Your configuration is significantly different from the recommended configuration in the documentation:

Package files are generated on the fly. There are no generator logs.

@epriestley I strictly change my conf file on documentation recommends and restart nginx, but this doesn't help.
Is there any way to load this files from CDN, for example?

Had exact same issue today on CentOS 7.1 box. I had changed the user that Nginx was running under to 'www-data'. Changing the ownership on /var/lib/nginx (chown -R www-data.www-data /var/lib/nginx) allowed nginx to write to its /tmp/ directory and fixed the broken CSS and JS.

Having a very similar problem - Phabricator runs fine with http. But when switching over to https, CSS, JS and images failed to load. What things I should look into?

I am running RHEL 7.1 in AWS. Nginx 1.8, php 5.4 with SELiux set to permissive.

@siliconrockstar, I didn't find the nginx tmp directory you mentioned, do I need to create one manually?