2

About two weeks ago we performed maintenance on our on-premise infrastructure in our office. Including applying updates to our Hyper-V-based virtual hosts, as well as our SAN. We also modified some settings on our six Cisco Aironet 1040 access points to increase throughput, as our previous IT team had set our APs to drastically throttle bandwidth - office staff has been complaining of slow wireless for quite some time, and frankly there was no good reason to continue throttling. This, however, is where it gets weird.

After completing maintenance and going home for the night - and foolishly thinking we hadn't run into issues - we arrived the next day to find that about half of our twenty-some-ought in-office personnel's PC's were getting a limited connectivity error. Believing it was an issue with stale IPs (as we had reset the DHCP server during our maintenance) we did the release\renew thing and confirmed a new lease was granted by the DHCP server. This worked for some, but 4 or 5 machines still cannot connect.

Over the last two weeks we have tried literally everything we can think of to solve the connectivity issue. I finally found a workaround for some of our Lenovo PCs with Broadcom wireless cards: replacing the Broadcom 802.11n driver with a Broadcom 802.11 b\g\n driver. Our users with Intel wireless cards (some Lenovo W-series, some Dell 6500\6400 series) are still pretty much stuck hard-wiring - which, as you can imagine, doesn't go over real well with users.

Moreover

  • You cannot ping another machine on the same network. Pinging cannot be seen by other machines on the network unless you ping 255.255.255.0

  • When the wireless card is disabled, then enabled the gateway appears in the ARP table. About 10-15 seconds later it is no longer there. We have tried adding a static entry to the ARP table - it doesn't help.

  • Performing a reset on the TCP\IP stack in windows doesn't help.

  • Re-imaging the machine does not make a difference - still limited connectivity. Nor does adding and reinstalling the same driver.

All in all I'm baffled as to why this is happening, and am curious if anyone else has run into something similar? Thanks in advance!

CRK
  • 35
  • 3
  • Also, I've tried swapping the wireless cards. If I change it a card of the same type it doesn't help - this to me says there is no MAC spoofing prevention security system bothering it on our switches. I can swap the cards for a broadcom wireless card, and apply the fix above. But this won't help when it comes time to deploy windows 10 in a couple months – CRK May 01 '15 at 18:14
  • I agree with @mfinni as it's surely a change. Can be too a bad AP that is showing bad symptom now (after a restart in example). For the dhcp, do you use balancing in your router/AP or HSRP ? Do you use ip-helper ? as it exist some cisco kb it it's the case as HSRP can block dhcp renewal. – yagmoth555 May 01 '15 at 18:37
  • No and No. DHCP renewal isn't the problem. The machine gets and IP and even identifies the network domain. The issue seems to be lower-level. If I install Wireshark on the machine I can watch it send out a thousand ARP requests for the gateway without getting a single response. Personally I think the WAPs are suspect, but I don't know what (exactly) is causing the problem. – CRK May 01 '15 at 18:53
  • Do you got a small soho AP? Try it with some PC in trouble. That will isolate if it's a AP config error. – yagmoth555 May 01 '15 at 19:03

1 Answers1

1

Focus on the changes you made that could have impacted this. You made changes to the Aironets, and now 1/2 your wireless PCs are impacted.

You should be smart and simply revert those changes immediately, to restore service to your users.

Then, you could test those changes one-by-one, preferably in a test environment, to see which of those changes (or combination of those changes) broke your wireless connectivity with your clients.

mfinni
  • 36,144
  • 4
  • 53
  • 86
  • We tried this. Literally the only setting we removed was a speed limit. However, we did try reverting the changes to no avail. Also, it is not half the machines; most of those originally effected we simply because of a stale DHCP lease. – CRK May 01 '15 at 18:11
  • Then another change also happened in the environment. – mfinni May 01 '15 at 18:26
  • I agree, something changed. I just don't know what and I am running out of places to look – CRK May 01 '15 at 18:50