[solved] - Traffic Graphs: Cannot get data about interface [2.0.2]

[jikjik101]

Cannot get data about interface ste0
This is on the traffic graphs and on the dashboard. I have two WANs and two LANs.
Sometimes, if i delete history in chrome, the traffic graphs will work okay for 5 or less minutes, then the problem occurs.
Sometimes, one of the interfaces shows the problem, sometimes all four of the interfaces.

i tried reinstalling the packages: squid, squidguard and lightsquid and also ln -s /usr/local/bin/perl5.12.4 /usr/bin/perl to no avail.
This only happened when i updated from 2.0.1 to 2.0.2-RELEASE (i386) built on Fri Dec 7 16:30:38 EST 2012.

It only occurs on Chrome but not on IE.
I dunno with FF.
When using chrome, i need to delete the history like every 5 minutes to refresh the graphs, but in IE it automatically reconnects the graph even it shows Server Error for a few seconds. In 2.0.1, everything was fine, been using Chrome 24/7 for the traffic graphs with no problem.

Thanks in advance for any help.

[tim.mcmanus]

I am experiencing the same issue.

2.0.2-RELEASE (amd64) built on Fri Dec 7 22:39:16 EST 2012 FreeBSD 8.1-RELEASE-p13

This is a clean install with my previous 2.0.1 settings imported in.

The issue is random and sporadic.  I have four interfaces on the dashboard that I am monitoring.  The WAN interface stopped graphing after a few minutes and then the WAN2 interface stopped graphing several minutes later.

I haven't tried other browsers or clearing the cache.  A scree refresh or logging in/out does not resolve the problem.

[jikjik101]

I attached my dashboard, one from Chrome and the other one from IE.
Both captured with a few seconds difference.
I dunno why chrome just show the errors in 2.0.2.
It was very fine before the update.
Yeah, I know most will say that it's just a small issue and just use IE.
But I prefer to use chrome as my browser.

Btw, the proxy state under lightsquid does not auto refresh after the update, neither in chrome nor in IE.

[slampert]

I can confirm this as well.  If I switch the physical interface it works for awhile but then eventually gets in this same state.

[jikjik101]

@slampert:

I can confirm this as well.  If I switch the physical interface it works for awhile but then eventually gets in this same state.

what browser are you using? I can only see this in Google Chrome.

[slampert]

@jikjik101:

@slampert:

I can confirm this as well.  If I switch the physical interface it works for awhile but then eventually gets in this same state.

what browser are you using? I can only see this in Google Chrome.

Sorry I thought it was implied.  I'm seeing it in Chrome as well.  Works fine in IE.

[tim.mcmanus]

I am using Safari 6.0.2 in Mac OS X 10.8.2.


[jikjik101]

@slampert: they[developers] might advise us to use IE  ;D
@tim.mcmanus: i believe you have no IE in Mac, so i don't know what they will advise you.  ;D

[cmb]

Uh, yeah we'd never in a million years recommend anyone use IE for any purpose whatsoever. A significant portion of us use Chrome, if there were a universal problem we'd have seen it long ago.

JimP testing this earlier, couldn't replicate. I've had all the graphs on 2 different systems going for upwards of a half hour, 0 problems, they're still updating as I'm writing this.

There was only one small change to any of the files related, this here:
https://github.com/bsdperimeter/pfsense/commit/c93b202d75e3df1aecbb25587e1a225a05fbcc9c

to fix IE being retarded. Given the scope and relevance of that change, it's improbable that's related. Maybe if you all have stale caches in your browser that Jim and I don't have. You can try going back to the previous revision of that file from here:
https://raw.github.com/bsdperimeter/pfsense/86e1405de40ca1ab1d7ce6726e1181b9ea32f359/usr/local/www/status_graph.php

and see what happens.

Going to have to get some feedback from those who are seeing the issue, since it's not an issue for any of us.

[cmb]

You guys have the Widescreen package installed by chance? Wondering if it's breaking stuff as it commonly does since it overwrites base system files.

Chrome is the #1 browser of visitors to this site, 42% of visits in the last month. Thousands of systems have been upgraded to 2.0.2 already. The fact that 3 people are reporting an issue lends to something unusual like a package being the case. There would be a massively long thread by now if it didn't work in Chrome in general.

[jikjik101]

@cmb:

You guys have the Widescreen package installed by chance? Wondering if it's breaking stuff as it commonly does since it overwrites base system files.

Chrome is the #1 browser of visitors to this site, 42% of visits in the last month. Thousands of systems have been upgraded to 2.0.2 already. The fact that 3 people are reporting an issue lends to something unusual like a package being the case. There would be a massively long thread by now if it didn't work in Chrome in general.

Packages: squid, squidguard, lightsquid, imspector, mailreport,sarg and vnstat2 only.
At first i was afraid that it is only my setup that was messed up so i didn't post the problem, but when 2 users confirmed my problem, it was a "sigh of relief" knowing that i didn't do something stupid during the upgrade or on my setup.

I will try to revert to the previous revision of the php file (but i don't know how, so i will still search the forum for how to upload that one  ???)
Thanks for your attention sir.

Update: I tried reverting the php file under /usr/local/www/status_graph.php and same problem happened.

[wallabybob]

Behaviour depends on version of Chrome?

[slampert]

I tried the old status_graph.php to no avail, but clearing my cache fixes it for awhile.  I turned off all my chrome extensions to rule that out as a cause.  I have only two packages installed, mtr-nox11 and nut.  My chrome version is 23.0.1271.97 m.

[tim.mcmanus]

I have nmap and pfBlocker installed, but no other packages.

I'm also working on a clean install of 2.0.2, however I did import my previous settings.

I'll try emptying my browser cache to see if it resolve it when it happens again.

EDIT:  Just lost one graph after keeping the Dashboard window open for 15 minutes.  Emptying the cache fixed the issue.

EDIT AGAIN:  Lost another graph, a different one this time and not sure how long it took to fail, but emptying Safari's cache and reloading the Dashboard page resolved the issue.

[cmb]

24 hours and counting, 8 live graphs on dashboard, all still working in Chrome. Latest version on Windows in that instance. Works on my Mac too though I haven't had them going on it for as long.

[tim.mcmanus]

When the graph failed I got the following error when I had Safari in developer mode.

Server error 500.

See enclosed screen shot.

EDIT:  Added a second image from an hour later.


[cmb]

is there anything in /tmp/php_errors.txt?

[tim.mcmanus]

@cmb:

is there anything in /tmp/php_errors.txt?

/tmp/php_errors.txt does not contain any data.

[tim.mcmanus]

Another interesting observation:  I disabled the caching in Safari and got three 500 errors.  However, since caching was disabled in the browser, the graphs never stopped working.


[cmb]

I remembered one other change, lighttpd was upgraded. Its upgrades have broken things in weird ways in the past, wonder if something in it is somehow causing 500s that only a seemingly very small % of users see. Anyone capable of extracting just the lighttpd files from the 2.0.1-Full-Update tgz and replacing those files, that would be a good test to see if that makes it go away. Just the lighttpd binary alone may be adequate.