0

I have an issue with my student computers - they can't connect to the internet.

Here's the weird part: I open a command prompt and ping our gateway server, and suddenly everything is fine (at least for a while). I don't know what the ping is doing to allow the connection to work.

I've tried ipconfig /flushdns and deleting and recreating routes, and it doesn't work. I thought maybe the ping was forcing machines to cache a dns entry, but it doesn't matter if I ping by IP or by dns name. Visiting the gateway server via a web browser also has no effect — a ping is the only thing I've found that helps.

No computers in my lab or admin subnets have this problem; it's just the student machines.s The only other distinguishing factor is that admin/lab machines are joined to our active directory domain, while student machines obviously are not.

Any idea what's going on here? It's got to be something stupid I should have spotted already.

Joel Coel
  • 12,910
  • 13
  • 61
  • 99
  • It is hard to help without more details. Strange problems can occur if there is an IP conflict. Also, check for a loop if you have multiple switches (without STP). – Khaled Dec 14 '10 at 08:03
  • Clarity on the network layout is required for more informative questions and answers. – Zayne S Halsall Dec 14 '10 at 12:36

5 Answers5

2

How about checking the router/switch ? If the switch's MAC table were confused or something, it wouldn't operate normally, usually it would be back operating if the switch were restarted or the client computer keeps pinging the gateway. Duplicate IP address is a possibility, the gateway's ARP table could be forced to accept client's IP if the client keeps pinging. I actually put a ping command into crontab in one of my servers so it would execute every 3 minutes, because there is another computer with duplicate IP that I couldn't locate.

YudhiWidyatama
  • 181
  • 4
  • This'd be first on my list given these symptoms – symcbean Dec 14 '10 at 14:12
  • This turned out to be the problem - there was an up and down fiber link between the switch that connects the gateway and the core switch on our network. I was fixated on the gateway machine because it had just recently (the day before) needed to be completely rebuilt. – Joel Coel Dec 14 '10 at 17:04
1

You mention arp, but don't post the arp table info. I suggest you pay attention to the arp table as you are able and unable to make network connections. It seems possible that there is some strange behaviour there.

How reliable / repeatable is this correlation?

Slartibartfast
  • 3,265
  • 17
  • 16
1

Could we start with do domain names resolve? If you ping google does it resolve an IP Address? If so what happens when you attempt to tracert google.co.uk ? at which point does it stop?

Has anything changed since this problem began? As mentioned above check for switching loops (If you dont have Spanning Tree Protocol) How are the student machines IP addressed? (Manual or DHCP) can the machines ping each other? (By name or IP)

I'm sure we will find the problem for you.

JamesK
  • 1,646
  • 11
  • 19
1

I believe this is ICMP-route-redirect.

Your router is injecting a "better" route on your hosts network segment.

  • Do a traceroute route to 8.8.8.8 and record your path.
  • Ping de default GW.
  • Do another traceroute to 8.8.8.8 and compare the results.

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a0080094702.shtml

voodooo
  • 254
  • 2
  • 6
0

You need to get a better understanding of the path of network devices and/or do a packet trace.

benc
  • 663
  • 1
  • 5
  • 13