500 - Internal Server Error

dominique.fournier

Hi,

/var/log/lighttpd.error.log contains a lot of errors :
2012-04-05 15:00:20: (mod_fastcgi.c.1732) connect failed: Connection refused on unix:/tmp/php-fastcgi.socket-1
2012-04-05 15:00:20: (mod_fastcgi.c.3025) backend died; we'll disable it for 1 seconds and send the request to another backend ins
tead: reconnects: 0 load: 1

Of course, it produce a lot of problems : bad sync beetween nodes, interface hanging, etc.

How can I debug this ? Do you reproduce the problem ?

It is the snapshot of Wed Apr 4 19:15:35 EDT 2012.
Thanks !

jimp

If you upgrade to a more recent snapshot, does this still happen? (You can use the console upgrade by URL function to load an updated snapshot)

databeestje

I had this happen after a gitsync.
Manually restarted the webui with 11.

dominique.fournier

Yes the problem persists on Mon Apr 16 04:08:01 EDT 2012 version.

jimp

We'll need more info about what you have configured there, or what you have happening with the GUI, since I can't reproduce this on any of my 2.1 systems, they all work fine.

dominique.fournier

I just install the last version on two Virtualbox VM. I define IPv4 and IPv6 address on one interface (WAN) for each. I define a CARP address on this WAN. I define a DNS server. I open the firewall, so for each VM can see the other, can ping google, update on PFSense.

Each time I click on a page the log is growing up.

What information can I give you ? Are you interested by the VirtualDisk of these machines ?

dominique.fournier

I try a lot of things :

Removing all IPv6 addresses and CARP
Increasing the Max Processes in Webconfigurator
Disabling IPv6 in Advanced menu
Try to add the name in /etc/hosts
Rebooting
Try to update to the updated version
Removing HA (pfsync and conf)
And no luck.

Just if it can help, the IPv4 for the PFsense is not resolvable.
When I increase the Max Processes to 4, the message change to unix:/tmp/php-fastcgi.socket-3 instead of unix:/tmp/php-fastcgi.socket-1

Just to be sure, I install a 2.0.1 on a new VM (no error detected), upgrade it to 2.1, and then the problem occurs.
I install a fresh 2.1 (on another VM) and the problem occurs too…

I saw an error message concerning a low memory (I use 128MB RAM VM, which is the default for VirtualBox). It disable APC. I try to put 1GB to VM and YEAH !! It works !!! (put 256MB works too and it is really quicker than without APC !!!)

jimp

OK, so if it's low memory, there are already a couple of tickets on http://redmine.pfsense.org/ that cover that type of issue. If you have that low memory you do not want to increase the number of GUI processes, it can only handle one.

dominique.fournier

Yes ! I put more memory and the problem is solved (for me). I think some guys will have the problem on embedded devices.

Thanks for your time !

jimp

Most embedded devices that are still in use have at least 256MB RAM, like most of the popular ALIX models. (There are a couple with 128MB but they are much less common)