We have an office network with a sonicwall TZ190 firewall and four daisy-chained switches, the first three 24 port 3com switches, the last one an older 12 port Cisco switch. (I know this isn't a great setup, next time I think I'll insist on 2 48 port switches). The issue I've been having is that for two days now, at around the same time each day, DHCP suddenly stops working, and the firewall is unreachable (can't be pinged) by a computer if I release its DHCP address.

If I assign a computer a static IP in this case, the computer is immediately connected with the router. Sometime later in the day, this problem magically disappears and DHCP works again.

I thought maybe it was an issue with a switch loop, although I can't find any loops, and I'm not sure if STP is active on the 3Com switches (but it should be right?). When I do a packet trace, I can see STP packets coming from the Cisco switch (which is at the end of the chain) but only from it. Honestly, I'm not really sure how to connect to the 3com switches. I've done Cisco switches from the command line with no problem, but I can't find the 3Com switches at the IP addresses listed on them (169.254.xxx.xxx). There's a CD with the detect utility but it doesn't find them either.

Not sure if this is relevant at all, but the Cisco switch is about 100m away from the other switches in a separate room. There are two hubs and a wireless AP (or a wireless router in AP mode) also connected to the network, but I haven't found any looping issues there either.

So, any ideas? Any info I've left out that you'd like?

Thanks mucho.

  • I also now realize that I need some kind of USB to serial adapter to access the 3Com switches, so at least I figured that part out. I think. – John Glass Mar 24 '11 at 7:09

I'm not sure if this helps but I had similar experience few years ago and what I found that my switch has a broadcast "storm control" feature that was blocking broadcasts time after time for no obvious reasons. I disabled the feature and everything worked fine ever since.

| improve this answer | |
  • Broadcast storm conrtol shouldn't inhibit the operation of DHCP unless it's misconfigured on the switch. – joeqwerty Mar 24 '11 at 12:14
  • Yes, it most likely was misconfigured but I saw no need for it in our network and just disabled it. – dtoubelis Mar 24 '11 at 16:23
  • No need to control a boradcast storm if it occurrs? – joeqwerty Mar 24 '11 at 17:10
  • @joeqwerty, you got me. I was lazy, I did not know what this feature is about but disabling it seemed natural and it fixed the problem ...so, now I'm passing the wisdom :-) – dtoubelis Mar 24 '11 at 17:39
  • Glad to help... if I did... – joeqwerty Mar 24 '11 at 23:30

If there ware a loop you'd be having more than just DHCP issues. My suggestion would be to look at the DHCP scope and make sure you have sufficient ip addresses to serve all of your clients.

In addition, when this issue occurrs have you checked the ip address on the client that you've released the DHCP ip address from? If the client has an ip address in the range to then the issue is that you don't have a big enough ip address range in your DHCP scope.

| improve this answer | |
  • Thanks for you answer. OK, I'd like to think this is the answer. But I have 198 IP addresses open and at the most 25 users. Also, when this problem occurs I can't even ping the router from the affected computers, and if I do a release and renew I can't get back on, even if I do it very quickly. – John Glass Mar 24 '11 at 22:10
  • Have you verified what, if any, ip address the client has when the problem occurs? I would suggest running ipconfig/all on an affected client and look at the IP information that's assigned to the client to check for a proper ip address, default gateway, etc. – joeqwerty Mar 24 '11 at 23:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.