Posted by qbnflaco on April 23, 2010 at 1:24pm
I just signed up with linode and used the stackscript for mercury. I booted up and logged in as root. I pinged the IP address of my VPS and after a bit got this...
Page Could Not Be Loaded
We're very sorry, but the page could not be loaded properly. This should be fixed very soon, and we apologize for any inconvenience.
Debug Info:
Status: 503
Response: Service Unavailable
XID: 1255939562
Varnish
It still seems to be loading endlessly. I checked the settings.php file and everything seems to be fine there. Anyone ran into this before?
Thanks!

Comments
2 things
Hi there. Two quick things to check:
Give that a shot and let us know where you end up. If that doesn't get you going, there is a file /root/stackscript.log that should offer some pointers.
Adding index.php worked for
Adding index.php worked for me! :-) Thanks!
Two obstacles and the solutions I discovered
I am also new to Mercury and have been installing it to my own KVM-based server running on Ubuntu Jaunty 9.04 (amd64). I experienced the same persistent problem (Varnish 503 error) over the course of several attempted installs. In my case, I discovered through Webmin that apache was not starting correctly (even though it had been appearing to start from the command line when I ran /etc/init.d/apache2 start). Thanks to prior posts in this form, I was able to trace the error to APC (the logs reported that insufficient memory was being allocated to APC). I checked the apc.ini configuration (located in /etc/php5/apache2/conf.d) and found that following completion of the init.sh script (located in /etc/mercury/init.sh), the apc.ini file contained the following setting: apc.shm_size=4096. I doubled the RAM allocated to my KVM virtual machine (from 1024M to 2048M) and changed the above-referenced setting (in apc.ini) to apc.shm_size=1024. I was then able to start Apache from Webmin and no longer received the 503 error from Varnish.
Then, in attempting to install Mercury (from the standard Drupal web installation interface), I could not connect with the MySQL database I had created (the error reported there was no /var/run/mysqld/mysqld.sock file). Again, thanks to the information in this form, I learned someone else had fixed this problem by editing SELinux. Since Ubuntu runs AppArmor in lieu of SELinux, I followed the excellent advice located at http://www.howtoforge.org/perfect-server-ubuntu-9.10-ispconfig-3-p3 to disable and remove AppArmor completely (as set forth in that article, the three commands to enter (in order) are as follows: 1) /etc/init.d/apparmor stop 2) update-rc.d -f apparmor remove 3) aptitude remove apparmor apparmor-utils. Once AppArmor was gone, I was able to run /etc/init.d/mysql and install Mercury without further difficulty.
Frankly, I don't know if these steps are optimal (if someone has better solutions, please correct me). Nevertheless, I wanted to add my experiences as noted above in the hope I can help someone.
Yep
There are lot of eccentricities with different distros and setups. The APC issue is likely a misreading of available memory on your system. We're working on figuring out why that happens in some cases (but not others).
The AppArmor issue is definitely something we're working on. There's a fix in trunk that should make it out to 1.0 and 1.1 soon.
Thanks for the bug report!
https://pantheon.io | http://www.chapterthree.com | https://www.outlandishjosh.com