Slow network, high ping time in my Thinkpad T60

Message

Feng Wang · #1 Post by **Feng Wang** » Wed Sep 13, 2006 4:33 am

Hi, all

The ethernet network adapter in my t60 is Intel PRO/1000 PL. Version of driver is 9.3.39.0 (Installed form Software Installer). We use a 100M switch to connect other computers. Normally the ping time is below 1ms. But if another computer ping my t60, the time is over 10ms, even hundreds of ms (See below). Doesn't anyone else trigger this problem?

My t60's ip is 202.197.20.52 and WFPC's ip is 202.197.20.51. They are all conneted to the same switch.
[wf@WFPC wf]$ ping 202.197.20.52
PING 202.197.20.52 (202.197.20.52) 56(84) bytes of data.
64 bytes from 202.197.20.52: icmp_seq=1 ttl=64 time=0.574 ms
64 bytes from 202.197.20.52: icmp_seq=2 ttl=64 time=50.5 ms
64 bytes from 202.197.20.52: icmp_seq=3 ttl=64 time=43.4 ms
64 bytes from 202.197.20.52: icmp_seq=4 ttl=64 time=33.5 ms
64 bytes from 202.197.20.52: icmp_seq=5 ttl=64 time=23.6 ms
64 bytes from 202.197.20.52: icmp_seq=6 ttl=64 time=13.6 ms
64 bytes from 202.197.20.52: icmp_seq=7 ttl=64 time=3.67 ms
64 bytes from 202.197.20.52: icmp_seq=8 ttl=64 time=431 ms
64 bytes from 202.197.20.52: icmp_seq=9 ttl=64 time=55.7 ms
64 bytes from 202.197.20.52: icmp_seq=10 ttl=64 time=45.4 ms
64 bytes from 202.197.20.52: icmp_seq=11 ttl=64 time=401 ms
64 bytes from 202.197.20.52: icmp_seq=12 ttl=64 time=391 ms
64 bytes from 202.197.20.52: icmp_seq=13 ttl=64 time=119 ms
64 bytes from 202.197.20.52: icmp_seq=14 ttl=64 time=16.0 ms
64 bytes from 202.197.20.52: icmp_seq=15 ttl=64 time=371 ms
64 bytes from 202.197.20.52: icmp_seq=16 ttl=64 time=361 ms
64 bytes from 202.197.20.52: icmp_seq=17 ttl=64 time=351 ms
64 bytes from 202.197.20.52: icmp_seq=18 ttl=64 time=341 ms
64 bytes from 202.197.20.52: icmp_seq=19 ttl=64 time=331 ms
64 bytes from 202.197.20.52: icmp_seq=20 ttl=64 time=41.0 ms

--- 202.197.20.52 ping statistics ---
20 packets transmitted, 20 received, 0% packet loss, time 19166ms
rtt min/avg/max/mdev = 0.574/171.510/431.237/167.271 ms

This really give me a headache. Usually I log in the server from my notebook to work. The high echo time makes my commands not execute in real-time.

Any suggestion is appreciated. TIA.

Feng Wang

claudeo · #2 Post by **claudeo** » Wed Sep 13, 2006 9:01 pm

Symantec firewall?

Feng Wang · #3 Post by **Feng Wang** » Wed Sep 13, 2006 9:35 pm

Thanks, claudeo.

claudeo wrote:Symantec firewall?

No. Even I closed the Windows XP's firewall.

#4 Post by **Kyocera** » Wed Sep 13, 2006 9:54 pm

Try pinging your own IP and also try 127.0.0.1, and what are the times if you ping your default gateway and your server IP address.

If you suspect it is your network adapter run PC doctor. Make sure your cable is good or even switch it just to make sure and check the wall jack you are plugging into.

Feng Wang · #5 Post by **Feng Wang** » Thu Sep 14, 2006 2:33 am

Kyocera wrote:Try pinging your own IP and also try 127.0.0.1, and what are the times if you ping your default gateway and your server IP address.

If you suspect it is your network adapter run PC doctor. Make sure your cable is good or even switch it just to make sure and check the wall jack you are plugging into.

The network can work but be slow. Ping to other computers (or gateway or itself) is working in order. But ping from others to t60 is very slow. I checked the cable and the jack and once changed the switch to a 10Mbps hub, but it still exists.

#6 Post by **Kyocera** » Thu Sep 14, 2006 5:50 am

Feng, another thing you could try is taking the network out of the loop and get another laptop sit it right next to the t60, connect with a short ethernet cable set up the tcp/ip for both and ping it that way, the only conclusion I could come to after that would be something wrong with the network card in the t60. If you are in a bind you could go pick up a pcmcia card with a rj45 connection and use it for now.

HTH.

#7 Post by **jdhurst** » Thu Sep 14, 2006 7:20 am

If you are on a switch, why does your TP have an outside address. If you are internal to a connection to the outside world via a switch, perhaps there is a TCP/IP or duplicate DHCP issue.
... JD Hurst

Feng Wang · #8 Post by **Feng Wang** » Thu Sep 14, 2006 8:58 am

jdhurst wrote:If you are on a switch, why does your TP have an outside address. If you are internal to a connection to the outside world via a switch, perhaps there is a TCP/IP or duplicate DHCP issue.
... JD Hurst

The switch does not connect to the Internet. We construct a LAN in the same subnet.
IP Address: 202.197.20.*, Subnet mask: 255.255.255.0

Feng Wang · #9 Post by **Feng Wang** » Thu Sep 14, 2006 9:22 am

Kyocera wrote:Feng, another thing you could try is taking the network out of the loop and get another laptop sit it right next to the t60, connect with a short ethernet cable set up the tcp/ip for both and ping it that way, the only conclusion I could come to after that would be something wrong with the network card in the t60. If you are in a bind you could go pick up a pcmcia card with a rj45 connection and use it for now. HTH.

Oh, Yes. I am just using an 10Mbps pcmcia card found in my old laptop used in the last century. It is faster than the GBits network card in the t60, at least on the performance of logging to my server and typing commands.

I just wonder if it is a software problem in t60. If not, I will require the seller to change one for me.

jon · #10 Post by **jon** » Thu Sep 14, 2006 3:40 pm

Hi Feng

I and a few others have this problem too. I've got it on my X60s. It seems to have nothing to do with network cables, switches, firewalls or the like. See if these links describe your problem:

http://forum.thinkpads.com/viewtopic.php?t=25307
http://www.hack.org/mc/freebsd-x60.html

I don't know what the solution is. I haven't returned my X60s over this issue (yet) as I'm usually on wireless. Occasionally pings over ethernet are fine until the next reboot but I can't reproduce it all the time. When it happens, ssh sessions feel like they're running over dial-up and Windows shares are slow to browse in Explorer. Rather annoying.

Let me know if you find a solution.

Wiz · #11 Post by **Wiz** » Thu Sep 14, 2006 5:16 pm

I had a similar issue with a gigabit ethernet card connected to a gigabit port on a cisco switch. I solved that problem by enable flowcontrol on the switch port. Before i enabled flow control it was really slow to copy files to and from that PC. If you don't have a switch with management or without that option you might try to disable flow control on the network card. I never checked the response time with ping though.
I have seen slow response time with ping because of power management, but only on wireless cards. Set the network card to max performance changed the ping response time and also increased the speed.

You could try as suggested by Kyocera and connect two T60's directly with a cable to see if the problem is still there and find out if the problem have to do with the switch or not.

Feng Wang · #12 Post by **Feng Wang** » Thu Sep 14, 2006 8:39 pm

jon wrote:Hi Feng

I and a few others have this problem too. I've got it on my X60s. It seems to have nothing to do with network cables, switches, firewalls or the like. See if these links describe your problem:

http://forum.thinkpads.com/viewtopic.php?t=25307
http://www.hack.org/mc/freebsd-x60.html

[snip]

Rather annoying.

Let me know if you find a solution.

jon, Thanks for your reply. From the two links, especially the second, it is exactly the same problem. I think this is a common problem. I tested all options in "Interrupt Throttling Rate" and forbid the "intelligent close power" (I don't know if it is translated correctly from my Chinese version winxp. It is the last second property of advanced network configuration), all didn't work. And I installed a Fedora Core 5 Linux on my t60, the symptom still existed.
But after using one or two hours, the ssh session works normally and the ping time is down to below 1ms. After rebooting t60, the problem jumps out again.
Yes, rather annoying.

Feng Wang · #13 Post by **Feng Wang** » Thu Sep 14, 2006 9:25 pm

Wiz wrote:I had a similar issue with a gigabit ethernet card connected to a gigabit port on a cisco switch. I solved that problem by enable flowcontrol on the switch port. Before i enabled flow control it was really slow to copy files to and from that PC. If you don't have a switch with management or without that option you might try to disable flow control on the network card. I never checked the response time with ping though.
I have seen slow response time with ping because of power management, but only on wireless cards. Set the network card to max performance changed the ping response time and also increased the speed.

You could try as suggested by Kyocera and connect two T60's directly with a cable to see if the problem is still there and find out if the problem have to do with the switch or not.

My switch has no flow control. And disabling that feature on t60 does not work.
We have only one t60. Another t43 works fine. I am finding a cable to connect them directly.

kai jiang · #14 Post by **kai jiang** » Fri Sep 15, 2006 2:54 am

Hi brethren
You can disable the "Deep Smart Power Down (DSPD)" function in network adapter properties.I have fixed this problem by this.

agarza · #15 Post by **agarza** » Fri Sep 15, 2006 8:00 am

Just google for foceASPI, install those drivers and the problem will be gone.

jon · #16 Post by **jon** » Fri Sep 15, 2006 5:19 pm

Neither of those things are the solution. Changing 'smart power down' (can't see the word 'deep' anywhere in my settings) makes no difference. As for the ASPI drivers - not sure why that would be relevant, but considering the problem happens in Linux & FreeBSD also, I doubt it's anything to do with Windows' ASPI layer.

Very confident suggestions though

snife · #17 Post by **snife** » Fri Sep 15, 2006 6:25 pm

Theres a problem in the Intel microcode on these chips - a fix will be released soon i think so keep an eye on the driver matrix.

Feng Wang · #18 Post by **Feng Wang** » Sat Sep 16, 2006 3:14 am

kai jiang wrote:Hi brethren
You can disable the "Deep Smart Power Down (DSPD)" function in network adapter properties.I have fixed this problem by this.

Hi, Kai

I reinstalled my network driver and found this property (深度智能关闭电源), and disabled it. But it does not work on my t60.

Feng Wang · #19 Post by **Feng Wang** » Sat Sep 16, 2006 3:30 am

snife wrote:Theres a problem in the Intel microcode on these chips - a fix will be released soon i think so keep an eye on the driver matrix.

Good news. Thank you, snife. Hope this is not a long term.

And the seller said to me that another t60 has the same issue. But there's one thing I was confused, don't they test their products before releasing them. IMHO this is an easy to find problem existing in so many t/x60 laptops.

snife · #20 Post by **snife** » Sat Sep 16, 2006 1:20 pm

I too have concerns over the testing, or lack of, that is being done on most products you buy these days including ThinkPads. I have even more concerns over Intels ability to write code as every driver they release to solve a problem at the moment seems to cause another 3 problems.

Feng Wang · #21 Post by **Feng Wang** » Mon Sep 18, 2006 8:12 pm

I find a solution. Just ping t60 from other machine.

After ONLY 10 to 20 minutes (sending about 1000 icmp packages), the latency will be normal and stable. See below:

[wf@WFPC wf]$ ping 202.197.20.52
...
64 bytes from 202.197.20.52: icmp_seq=1150 ttl=64 time=175 ms
64 bytes from 202.197.20.52: icmp_seq=1151 ttl=64 time=446 ms
64 bytes from 202.197.20.52: icmp_seq=1152 ttl=64 time=436 ms
64 bytes from 202.197.20.52: icmp_seq=1153 ttl=64 time=426 ms
64 bytes from 202.197.20.52: icmp_seq=1154 ttl=64 time=416 ms
64 bytes from 202.197.20.52: icmp_seq=1155 ttl=64 time=406 ms
64 bytes from 202.197.20.52: icmp_seq=1156 ttl=64 time=139 ms
64 bytes from 202.197.20.52: icmp_seq=1157 ttl=64 time=129 ms
64 bytes from 202.197.20.52: icmp_seq=1158 ttl=64 time=119 ms
64 bytes from 202.197.20.52: icmp_seq=1159 ttl=64 time=366 ms
64 bytes from 202.197.20.52: icmp_seq=1160 ttl=64 time=0.462 ms
64 bytes from 202.197.20.52: icmp_seq=1161 ttl=64 time=0.371 ms
64 bytes from 202.197.20.52: icmp_seq=1162 ttl=64 time=0.358 ms
64 bytes from 202.197.20.52: icmp_seq=1163 ttl=64 time=0.371 ms
64 bytes from 202.197.20.52: icmp_seq=1164 ttl=64 time=0.377 ms
64 bytes from 202.197.20.52: icmp_seq=1165 ttl=64 time=0.373 ms
64 bytes from 202.197.20.52: icmp_seq=1166 ttl=64 time=0.372 ms
64 bytes from 202.197.20.52: icmp_seq=1167 ttl=64 time=0.369 ms
...

Hope it learns more quickly when the next version driver is released.

jon · #22 Post by **jon** » Mon Sep 18, 2006 8:47 pm

But does that last beyond a reboot? I've noticed that it gets better after doing that for a while (try ping flooding it with ping -f instead) but it doesn't last.

Feng Wang · #23 Post by **Feng Wang** » Mon Sep 18, 2006 9:44 pm

jon wrote:But does that last beyond a reboot?

No. After rebooting we need ping again.

Nightwolf · #24 Post by **Nightwolf** » Tue Sep 26, 2006 6:00 pm

This is my configuration: T60 as FTP server<-> 10/100 Hub <-> other PC as FTP client
Both get their IP from the router that's connected to the hub.

Also I've sent pings till it was normal, the FTP client disconnected all the time. Right after the disconnects I got "host not found" messages when I tried to ping the T60. About 1-2 minutes later it worked again but only for about 4-5 minutes, then I was disconnected another time (and so on...). What I wonder about is: I downloaded something from the internet (HTTP and FTP) through the whole night without any disconnect. That's the weirdest problem I've ever had...

jon · #25 Post by **jon** » Fri Sep 29, 2006 10:23 pm

Anyone tried the latest PRO2KXP.EXE (version 11.1) from the Intel
website? I've been using it for a few days and ping times have been ok ever since I installed it.

http://downloadfinder.intel.com/scripts ... 1&lang=eng

#26 Post by **Kyocera** » Sun Oct 01, 2006 6:36 pm

Wondering if anyone has tried calling Tech support about this issue. After experiencing some strange connectivity issues with my machine I decided to check the ping times and found that I too have this problem. Jon, that driver did not help mine, glad it worked for you. I'm going to try and call tomorrow and report back if I get time.

Nightwolf · #27 Post by **Nightwolf** » Mon Oct 02, 2006 10:21 am

jon, do you have a PRO/1000 PL? I can't get the driver installed correctly. There's always a ! in the device manager and the describtion is code 10 (which means the device could not be started). Also the PRO/1000 PL is not in the list of of valid products for this driver (see http://downloadfinder.intel.com/scripts ... =&lang=eng).

#28 Post by **Kyocera** » Mon Oct 02, 2006 12:49 pm

Nightwolf, did you go to the Intel folder in C: and run the program from there? That's the way it worked for me but it still did not help the ping times.

Nightwolf · #29 Post by **Nightwolf** » Mon Oct 02, 2006 5:34 pm

Kyocera wrote:Nightwolf, did you go to the Intel folder in C: and run the program from there? That's the way it worked for me but it still did not help the ping times.

That was the first I tried. I also tried to uninstall the driver first and to install the driver manually but this doesn't change anything.

jon · #30 Post by **jon** » Mon Oct 02, 2006 8:19 pm

Nightwolf wrote:jon, do you have a PRO/1000 PL?

In Windows it's listed as that, yeah. But the chip is an 82573L.

I just tried my machine again from a cold boot and it's still pinging fine. In about a week of use I haven't been able to reproduce the bad pings, that's the best I can say. Can't think of anything else I changed other than upgrading to the latest BIOS.

Haven't called tech support myself. Let us know if you do Kyocera.

Slow network, high ping time in my Thinkpad T60

Slow network, high ping time in my Thinkpad T60

Who is online