Web UI crash every 5 minutes or so
-
I had a few UI issues in 2.0.2. I've upgraded (manual firmware update) to the most recent 2.0.3 release (Feb. 9) and that resolved all of my UI issues. You might want to do that. Download the latest .tgz image and manually update your pfSense firmware. IMHO it's very stable and I have a feeling it'll resolve your issues.
-
If you can still SSH into the box, what happens if you restart the web gui (menu option 11)?
If you run top does it show some process using all the cpu cycles?Steve
-
Upgrade to 2.0.3. We would have released it already but we're awaiting the "any day now" FreeBSD security advisory for the recent openssl issue.
-
Thanks for your answers. I had already tried restarting the webgui with console option #11, without any change.
I'm upgrading to 2.0.3 and I'll tell you if it works :) -
Tried v2.0.3, still crashes :(
I'm using a ESXi 5.1 based VM. Host is using a Xeon E5620 CPU (2.40GHz).
-
Anything in the system logs? Also check the lighttpd logs too.
clog /var/log/lighttpd.error.log
and
clog /var/log/lighttpd.log
-
Anything in the logs… This is quite bad.
When i'm connected to pfSense whith SSH, it seems like it takes a longer time to crash (I managed to keep control of the UI for 43 minutes yesterday).
-
Could you give a bit more detail on what you mean by "crash"? the page doesn't complete loading? you get an unexpected error report? (If so, what is the report?)
What happens when you close and restart the web browser?
Can you reliably get it to "crash" by going to a particular page? If so, what page?
Please post the output of pfSense shell command```
top -S -H -d 2 -
I don't know if this is specific to the version of pFsense I am running or is a general problem. I don't know if it related to the problem reported in this thread.
I went to Diagnostics -> Command Prompt, typed a command in the Command box and hit the Enter key. The web page is "stalled", the browser reporting (after some minutes) Waiting for … (the name of my pfSense box).
I clicked on the browser "Abort" button then attempted to go to the home page of my pfSense box and the browser again reported Waiting for …
I am using Chromium. I then used Firefox (without exiting from Chromium) and connected to the home page of my pfSense box. My pfSense box is running
2.0.1-RELEASE (i386)
built on Mon Dec 12 17:53:52 EST 2011
FreeBSD 8.1-RELEASE-p4as reported on the home page.
-
I went to Diagnostics -> Command Prompt, typed a command in the Command box and hit the Enter key. The web page is "stalled", the browser reporting (after some minutes) Waiting for … (the name of my pfSense box).
If you run a command there that never ends, like "ping google.com", that's the expected result.
-
The command was```
/etc/rc.banner -
Could you give a bit more detail on what you mean by "crash"? the page doesn't complete loading? you get an unexpected error report? (If so, what is the report?)
No HTTPS response is returned to my browser (not even a 4XX or 5XX code). Nothing. Chrome says "This webpage is not available". You don't say…
I'll try HTTP.What happens when you close and restart the web browser?
Nothing. Tried rebooting, switching to other browsers, it didn't work… This is a server-side error.
Can you reliably get it to "crash" by going to a particular page? If so, what page?
Nope, it's a random crash. Sometime i'm just managing my firewall rules, then I take a brake and when I come back, bam ! Crash.
Please post the output of pfSense shell command```
top -S -H -d 2
-
Edit : Holy **** ! It still crashes using HTTP : "No data received".
Just to let you know : i reverted to HTTP (Console –> Set interface(s) IP address --> x.x.x.x /x --> DHCP ? (n) --> Revert to HTTP ? (y)).
Web gui came back without needing a reboot ! This is great. Hope it won't crash anymore, but this already seems to be an improvement.
HTTPS isn't usefull at all in my case (separate network for gui), so i'll just keep HTTP and let you know if it stays up. :) -
Anything in the system logs? Also check the lighttpd logs too.
clog /var/log/lighttpd.error.log
and
clog /var/log/lighttpd.log
I never saw these logs posted. These logs are the service logs and will most definitely determine why this particular service is crashing. Please post them.
-
As I already said, those files contain absolutely nothing.
As a matter of fact, the lighttpd.error.log file doesn't even exists. -
It seems very odd that neither lighttpd or php are generating any errors. And the only thing you're having an issue with is the webUI. The rest of the installation works okay?
-
This is very strange, indeed. As long as I know, everything is working great apart from the web GUI (I don't think I'm using 100% of pfSense's functionnalities, but I haven't noticed any other problem).
My PPTP VPN is OK, my firewall rules run as expected, routing is working great…
I'm experiencing this problem on ONE COMPUTER ONLY.
I always run pfSense as a VM on top of ESXi 5.1. There's only ONE HOST doing this. I tried removing the VM and starting a new one from scratch, with a fine ISO (checked checksum). It does exactly the same thing. Could hardware be the problem ? -
Are you running snort?
It's seems strange that port 80 works and then becomes inaccessible. If the webUI were unaccessible, I would assume something would be generating an error log. It could be hardware, it could be ESXi, and it could be something running on the pfSense installation. I've had snort block me out of my own LAN before and other configuration issues that I created on my own.
One last thing, check the firewall logs on the pfSense box and see if your IP is getting blocked on port 80 for some reason.
-
I tried to eliminate every "custom" thing that could fail.
Basically, it's just my computer and the pfSense VM. Nothing between us, no firewall rule blocking or rejecting traffic.
The web gui comes back for a few minutes when I reboot the pfSense VM. -
It's strange I don't have a lighttpd.error.log.
All the logs are stored in /var/log right ?
Is there anything else I could check to try to find the error ? For the moment, I don't have a single clue and I bet it's even harder for you trying to diagnosticate the problem.
Thanks a lot.
Edit : All my VMWare ESXi interfaces are "flexible". I added one more interface with the "E1000" type and put it on the same network as my web UI interface.
I've been able to access the web ui with this second interface, although it was crashed (again) on the first one.
So could this possibly be a problem linked to the type of interface within ESXi ?
I'll wait and see if this second interface crashes too…
