Very slow internet browsing

firefox · Apr 15, 2014, 5:59 AM

Good Morning {in my area}

Now Morning
And out of 10 computers are connected and surf them {brothers sisters parents}
Only my computer works now
And the graph shows a lot of connections

browser Open with two tabs
One of the dashboard
Second of this page with the message

I guess there is a log in which all registered
Where can I see it.

firefox · Apr 15, 2014, 6:00 AM

Addition
Now I checked the Internet link with Mlab Shaperprobe

http://www.measurementlab.net/tools/shaperprobe

Although the dashboard
Says there are no packet loss
The test stops in the middle
And gives a message

"measurement aborted due High packet loss rate"

hagak · Apr 16, 2014, 3:23 PM

I have narrowed my issues down to just IPv6 I believe.

My IPv4 gateway appears to keep a solid connection with a ~10ms latency, however my IPv6 gateway varies from 10ms - 250ms. Any ideas what could have caused this issue with the most recent patch?

Prior to the patch I did not have this issue.

stephenw10 · Apr 16, 2014, 10:50 PM

Another user reported similar symptoms when using Intel NICs and some tweaks applied. Are any of you running similar setups?

Steve

firefox · Apr 17, 2014, 5:09 AM

I do not know what network cards I have
But I know that I did not changed their settings

Is there a command that
Show me the computer's hardware

ESPNSTI · Apr 17, 2014, 6:13 AM

@firefox:

Is there a command that
Show me the computer's hardware

pciconf -lv

It won't spell out the nic brand or model, you'll probably need to Google the chip number.

You can also look at this file to get the driver version, but that won't list the specific nic model:

/var/log/dmesg.boot

stephenw10 · Apr 17, 2014, 12:23 PM

Well if you haven't added any tweaks in loader.conf.local (for which you'd have to know what hardware you're running) then this is a different problem. Or at least a different solution.
You can tell what NICs you have by what drivers are attached to them. Intel NICs will usually use the em or igb drivers so your interfaces will be em0, em1 etc. Beyond that you need the PCI info for more detail but here the drivers are enough.

Steve

firefox · Apr 17, 2014, 12:51 PM

pciconf -lv

give me this

$ pciconf -lv
hostb0@pci0:0:0:0: class=0x060000 card=0xffffffff chip=0x25608086 rev=0x01 hdr=0x00
class = bridge
subclass = HOST-PCI
vgapci0@pci0:0:2:0: class=0x030000 card=0x02671014 chip=0x25628086 rev=0x01 hdr=0x00
class = display
subclass = VGA
uhci0@pci0:0:29:0: class=0x0c0300 card=0x02671014 chip=0x24c28086 rev=0x01 hdr=0x00
class = serial bus
subclass = USB
uhci1@pci0:0:29:1: class=0x0c0300 card=0x02671014 chip=0x24c48086 rev=0x01 hdr=0x00
class = serial bus
subclass = USB
uhci2@pci0:0:29:2: class=0x0c0300 card=0x02671014 chip=0x24c78086 rev=0x01 hdr=0x00
class = serial bus
subclass = USB
ehci0@pci0:0:29:7: class=0x0c0320 card=0x02671014 chip=0x24cd8086 rev=0x01 hdr=0x00
class = serial bus
subclass = USB
pcib1@pci0:0:30:0: class=0x060400 card=0x00000000 chip=0x244e8086 rev=0x81 hdr=0x01
class = bridge
subclass = PCI-PCI
isab0@pci0:0:31:0: class=0x060100 card=0x00000000 chip=0x24c08086 rev=0x01 hdr=0x00
class = bridge
subclass = PCI-ISA
atapci0@pci0:0:31:1: class=0x01018a card=0x02671014 chip=0x24cb8086 rev=0x01 hdr=0x00
class = mass storage
subclass = ATA
none0@pci0:0:31:3: class=0x0c0500 card=0x02671014 chip=0x24c38086 rev=0x01 hdr=0x00
class = serial bus
subclass = SMBus
none1@pci0:0:31:5: class=0x040100 card=0x02671014 chip=0x24c58086 rev=0x01 hdr=0x00
class = multimedia
subclass = audio
fxp0@pci0:2:8:0: class=0x020000 card=0x02671014 chip=0x10398086 rev=0x81 hdr=0x00
class = network
subclass = ethernet
ath0@pci0:2:9:0: class=0x020000 card=0x2051168c chip=0x0013168c rev=0x01 hdr=0x00
class = network
subclass = ethernet
dc0@pci0:2:10:0: class=0x020000 card=0x434e4554 chip=0x91021282 rev=0x31 hdr=0x00
class = network
subclass = ethernet
fxp1@pci0:2:12:0: class=0x020000 card=0x000c8086 chip=0x12298086 rev=0x08 hdr=0x00
class = network
subclass = ethernet

ESPNSTI · Apr 17, 2014, 11:13 PM

Possibly not 100% correct, but the nics might be these:

ath0 - Atheros Communications Inc. - AR5212, AR5213 802.11a/b/g Wireless Adapter
or 
ath0 - Atheros Communications Inc. - 802.11a/b/g Wireless Adapter (AR2312)

dco - Davicom Semiconductor Inc. - 10/100 Mbps Fast Ethernet Controller (DM9102/A/AF)

fxp0 - Intel Corporation - 82801DB PRO/100 VE (LOM) Ethernet Controller

fxp1 - Intel Corporation - 82550/1/7/8/9 EtherExpress PRO/100(B) Ethernet Adapter

firefox · Apr 18, 2014, 10:17 AM

Is there a way to know if Intel cards have a problem with pfsense

With the latest version Anyhow

stephenw10 · Apr 18, 2014, 12:22 PM

Just to clarify this, there is not any known bug in the Intel driver or card setup (that I know about at least). I mentioned this only because the symptoms described were very similar to those I had been reading about in another thread.

https://forum.pfsense.org/index.php?topic=74942.0

In that instance the high latency was seemingly solved by removing the loader.conf.local setting limiting igb to 1 queue.

Steve

firefox · Apr 19, 2014, 4:51 AM

I know I understood it the first time
My questions were just to rule out a network card problem

What could it be
why is the latency go crazy
from 10 to 300
why all internet browsing so slow
Except from the upgrades i did not changed anything

stephenw10 · Apr 19, 2014, 9:41 AM

OK. I just wanted to be sure I hadn't mislead anyone.
Your interfaces all appear to be older cards, 2 fxp NICs and one dc. Are you seeing latency problems across all of them?
The driver changes I was talking about were for the newer Intel drivers (em. igb, ixgb) for gigabit NICs and they went into 2.1.1. I don't beleive there were any driver changes between 2.1.1 and 2.1.2.

@firefox:

The temperature of the CPU is also higher than normal

This would indicate that the CPU is working hard on something which could be causing the latency. What do you see if you run 'top -SH'?

Steve

firefox · Apr 19, 2014, 4:08 PM

top -SH

here

$ top -SH
last pid: 62178;  load averages:  0.07,  0.14,  0.10  up 8+11:39:44    19:01:24
269 processes: 3 running, 248 sleeping, 18 waiting

Mem: 463M Active, 52M Inact, 143M Wired, 16M Cache, 84M Buf, 50M Free
Swap: 2048M Total, 49M Used, 1999M Free, 2% Inuse

  PID USERNAME      PRI NICE   SIZE    RES STATE    TIME   WCPU COMMAND
   11 root          171 ki31     0K     8K RUN    166.9H 96.39% idle
80016 root           44    0 58740K 13392K piperd   0:06  0.39% php
   12 root          -68    -     0K   144K WAIT   132:21  0.00% intr{irq20: fxp0 fxp1}
    0 root          -68    0     0K    88K -      109:12  0.00% kernel{ath0 taskq}
   12 root          -32    -     0K   144K WAIT    96:06  0.00% intr{swi4: clock}
    0 root          -68    0     0K    88K -       40:43  0.00% kernel{dummynet}
79394 root           44    0   262M   230M select  17:08  0.00% clamd{clamd}
56758 root           44    0  8004K  4616K kqread   8:06  0.00% lighttpd
   12 root          -44    -     0K   144K WAIT     7:06  0.00% intr{swi1: netisr 0}
  257 root           76   20  3352K     0K kqread   6:50  0.00% <check_reload_stat>
   14 root          -16    -     0K     8K -        5:03  0.00% yarrow
 9389 proxy          44    0 43124K 36700K kqread   3:57  0.00% squid
23208 root           64   20  3264K  1104K select   3:43  0.00% apinger
   22 root           20    -     0K     8K syncer   3:42  0.00% syncer
75378 root           44    0  3264K   804K piperd   3:24  0.00% logger
   12 root          -68    -     0K   144K WAIT     2:48  0.00% intr{irq22: dc0}
68925 root           44    0  3416K  1144K select   2:35  0.00% syslogd
    0 root          -68    0     0K    88K -        2:33  0.00% kernel{ath0 taskq}</check_reload_stat>

Currently the temperature is about 53 ° C
and latency 18
It reached 56 ° C

Before the upgrade it was from 30 to 36 maybe 40 in an exceptional case
And the latency Was up to 30 maximum

stephenw10 · Apr 20, 2014, 1:15 AM

If you run it from the console you will see the line detailing how the usage breaks down, like this:

[2.1.2-RELEASE][root@pfsense.fire.box]/root(1): top -SH

last pid: 99751;  load averages:  0.02,  0.02,  0.00    up 2+08:39:49  01:57:30
107 processes: 3 running, 88 sleeping, 16 waiting
CPU:  0.0% user,  0.0% nice,  0.0% system,  0.0% interrupt,  100% idle
Mem: 61M Active, 47M Inact, 103M Wired, 1248K Cache, 59M Buf, 269M Free
Swap: 

  PID USERNAME PRI NICE   SIZE    RES STATE    TIME   WCPU COMMAND
   10 root     171 ki31     0K     8K RUN     55.1H 100.00% idle
   11 root     -32    -     0K   128K WAIT     8:00  0.00% intr{swi4: clock}
  286 root      76   20  3352K  1192K kqread   6:55  0.00% check_reload_status
   11 root     -68    -     0K   128K WAIT     5:46  0.00% intr{irq18: em0 ath0
   11 root     -68    -     0K   128K WAIT     4:54  0.00% intr{irq17: fxp2 fxp

You have more than twice the processes I do, though I'm not running many packages on the above box.

A couple of things that have caused similar symptoms in the past come to mind. If you have any sort of CPU frequency scaling enabled and it's not quick enough responding to the load it cam start to appear like a large load. Doesn't really explain why it's running hot though.
The other thing is ASPM which can cause problems similar to this. Not sure if that's PCIe only though. Worth checking if it's an option in your BIOS and disabling it if it's there.

Steve

adam65535 · Apr 20, 2014, 1:22 AM

@ESPNSTI:

pciconf -lv
It won't spell out the nic brand or model, you'll probably need to Google the chip number.

I always add the file that pciconf can use to translate device ids to descriptive device names on all my pfsense installs so that the list is much easier to read. I really wish pfsense would include that file by default. It is very useful but it is about 400 KBytes.

/usr/share/misc/pci_vendors

I just do a search for that file for whatever version of freebsd pfsense is running and manually put it in place. pciconf is so much easier to read after that.

This…


em2@pci0:4:0:0:	class=0x020000 card=0x060a15d9 chip=0x10d38086 rev=0x00 hdr=0x00
    class      = network
    subclass   = ethernet

becomes this…


em2@pci0:4:0:0:	class=0x020000 card=0x060a15d9 chip=0x10d38086 rev=0x00 hdr=0x00
    vendor     = 'Intel Corporation'
    device     = 'Intel 82574L Gigabit Ethernet Controller (82574L)'
    class      = network
    subclass   = ethernet

firefox · Apr 22, 2014, 12:10 PM

The other thing is ASPM which can cause problems similar to this

I do not have this Option
I checked

now the latency in 325
Heat 50 °

Could be something on the network makes this
Perhaps improper connection of the network jack
Or improper cable

Do not know what to check

stephenw10 · Apr 22, 2014, 1:29 PM

If it were a hardware problem I would expect to see a load of errors in Ststus: Interfaces:

You aren't seeing a high interrupt load then if you run 'top -SH' at the console (or via ssh)?

Steve

firefox · Apr 22, 2014, 5:41 PM

What came out of this command
I posted earlier post

stephenw10 · Apr 24, 2014, 6:00 PM

If you look at my output from top -SH above I have an extra line the breaks down the usage by type. If you run it fro the console, rather then via the webgui, and allow it to go a few cycles you'll see that. It probably doesn't contain anything useful but at this point it's worth checking anything. :)

Steve