Strange network issue

I've been having issues recently with my internal network, and I'll attempt to give as much information as possible. The issue is my internet connection (roughly); I continue to lose the ability to load webpages, but its seemingly random, as the connection will come back with time. I'm using windows XP currently, and the adapter says connected, although sometimes pages online will not load. The strange thing is, I have another computer connected to the same switch (connected to the same router), and the other computer works just fine. Which would normally lead me to believe that it is the computer that is to blame, but I have another computer running linux that is connected directly to the router that seems to work intermittently, but has issues (seemingly) at the same time as my XP machine. Now it could just be coincidental, but I can't seem to come to any specific conclusion as to why it is occurring. Another strange issue is when I try doing the repair connection on the XP machine it fails at clearing the DNS cache. This made me try using IP addresses instead of domain names, which did not work either. I simply cannot figure out what is causing this.

It seems like it could be an issue with the DNS client service, which I needed to restart. But that doesn't necessarily solve the problem, as for some reason the service is having issues starting and stopping.

Wired or wireless?

When you have connection issues, what happens if you release and then renew the IP on either machine? Can you ping Google from either machine?

It seems as though I can ping google, but pages will not load. All my computers are wired. Releasing and renewing seems like it might be doing something, as usually google's home page will load after the connection is reestablished, but then not much will work after that. But actually, instead of renewing the IP I've been disabling and re-enabling the device, as I believe that does renew the IP.

ipconfig /flushdns

Try that on the xp machine.

Also maybe try and lower the MTU settings on the router

http://www.butterscotch.com/tutorial...e-MTU-Settings

MLF

Try an alternative DNS server. Level3 has public DNS servers at:

4.2.2.1
4.2.2.2

Google's are at:

8.8.8.8
8.8.4.4

OpenDNS:

208.67.222.222
208.67.220.220

Any of those should be reliable.

I flushed the DNS, lowered the MTU to 1000 from 1500, and set my DNS servers to OpenDNS, and I am still having the same issue. Any other ideas?

Hardware? Got an extra NIC around?

Could it be an ISP issue? Are you dropping packets?

I often run continuous pings against www.yahoo.com, sometimes for
days at a time, to gauge packet loss. You could run a continuous ping
against www.google.com too.

Trying to gauge packet loss is one of the first things I do on a troubled
network

ping -t www.yahoo.com

Also, what are you running for a gateway? I've had recent problems on a
free wifi network I admin when I set udp filtering a little too low and the users
were unable to connect to a DNS server. Oops. :o

I'm using a netgear FVS318 as my router. I haven't pinged google for long periods of time, but every time I had issues with pages loading I got 0% packet loss (with only 4 pings, as thats the default in xp), but I will try a continual ping and get back to you.

How long has that Netgear FVS318 been around? I have a "pull" from a
client's office that flaked out similarly to what you described after 3 or 4
years. I did update the firmware and continue to use it on a non-essential
network segment (my bench, easily swapped out), but I can't say I trust
it.

You might try a different gateway...

p.s. -- the standard 4 pings isn't going to tell you much. Ping is your friend. :)

i just ran 654 pings and only one was lost. I can keep going if you would like. I think the router is from around 2004, but I'm not exactly sure. Would it be worth trying to update the firmware?

I wouldn't update the firmware if I didn't feel I had to, or if I didn't have a back up unit.

One packet lost in 654 is perfectly acceptable. Anytime you get more than 1% packet loss and things go downhill. Believe me, we've got RoadRunner Business Class here and every 3 or 4 months we get a weekend of 2-5% packet loss.

Try running ping for extended periods. Before RoadRunner ran new cable on the poles nearby, we'd get much higher packet loss on a cold snowy day after dark but during the daytime things were relatively better.

Things got a little better with the new line, but they still get server issues in this area at times. Also, run ping against your internal network ip's for a benchmark too. RR's line techs routinely use ping, they even have special devices they plug the coax cable in to test.

Ping the chit out of everything and see what, if anything, is getting dropped. And don't worry, no one's ever showed up at my door or written me nasty letters for running ping (imcp packets) against their ip's.

HTH, your rogue admin

p.s. -- you can run multiple instances of ping on a given PC.

p.p.s. -- see if a friend will lend you a router. That FVS318 is old.

In my case the DHCP service caused the trouble, try to assign the addresses manually.

Another possible cause may be malfunctioning proxy server, are you using any?

And one last thing, perform "tracert" to google from all your devices, the output can be very helpful in order to find your problem.

I don't see how the gateway as a whole could be a problem, as it serves connections to at least two other computers just fine. Is there any reasoning behind this, or just because its old?

Well, you have two computers going through a switch to the router - one works fine all the time, the other loses connectivity. A third computer, hooked straight into the router, loses connectivity seemingly at the same time the one behind the switch does.

Very interesting.

The first thing I would do would be to plug all three directly into the router and see if they continue to experience problems. If not, then replace the switch.

If the problems do continue, then replace the router.

I'm assuming you've already tried resetting the router to factory settings right? Sometimes if mine starts acting weird, I'll just do a hard reset and reconfigure the whole thing and voila.

Check your host file, and otherwise it sounds like a ipstack issue.

open command prompt and type "netsh winsock reset" without the quotation marks. Providing your DNS is correct and the issue is not browser related it should fix the issue.

Have you tried a different browser?


CTO

Quote:

---

Originally Posted by CybertecOne

Check your host file, and otherwise it sounds like a ipstack issue.

open command prompt and type "netsh winsock reset" without the quotation marks. Providing your DNS is correct and the issue is not browser related it should fix the issue.

Have you tried a different browser?

---

The fact that he has issues with the machine behind a switch and a second computer connected to the router lead me to believe it is a hardware error or conflict.

Well this same machine (one of the ones with network issues) is the one I am installing the new linux server distro, so I'll run that for a few days and see if it's a software issue with XP.

Quote:

---

The fact that he has issues with the machine behind a switch and a second computer connected to the router lead me to believe it is a hardware error or conflict.

---

I only skimmed the post, but noticed it had not been said :p


CTO - interested to see how the new OS goes.

At this point I'm going to guess that it's XP. I've had continued service in Debian and it's been running continuously since I installed it. I'm guessing its a driver problem for my network card. I'll have to continue the investigation. :rolleyes:

The individual ports themselves, where the RJ45's plug in, will go bad.
On the FVS318, there's one WAN port and 8 LAN ports. Obviously the
WAN port's dedicated, but have you tried switching the cable from the
problem PC to another LAN port? Again, that FVS318 is old and this
stuff gets flaky. You can try that on the switch, too.

HTH

I think brokencrow maybe onto something there

Many moons ago I had one machine with continued network issues and disconnects...changed the nic, cable to the wall and finally reinstalled the OS. Still had the issues. Finally called someone to test the punch down port (I now have a tester)....yeap that was the problem. Punched down to another port...voila all fixed and lesson learned. I now have a port tester and a very long cable to connect to another port to confirm ;)

MLF

Not that this scenario fits your circumstances, but just throwing it out there as the issue is similar.

Within a network a terminal server would allow rdp connections, but once connected the internet did not work, sometimes you would get a page but very rare. From a terminal session the internet also failed. Without going into all the troubleshooting, the issue was the NIC hardware settings (NIC properties), and the conflict was between a switch and the router MTU settings.

So, switch and router are pre-existing and work fine. All other servers work fine. Terminal server network connectivity all great, rdp, network shares etc - except the internet. The packets traveling from the NIC into switch which then passes to the router (all other network traffic/testing does not get to the router).

From vague memory it had something to do with the intel NIC "adaptive" technology. disabling these settings resolved the issue. However, setting the switch MTU and router MTU to the same value also proved to be a more permanent solution. (i forgot to mention switch MTU and router MTU had different values.)


[it took 2 days to figure out many years ago]


CTO

I think that even though there were multiple computers having issues, it was just a coincidence. I've been running a Debian distro 24/7 with a perfectly fine connection since the install. I'm going to go online and try to find updates for the drivers for XP. If and when I boot back into it, haha.