Deployed the Nanostation as an AP and having weird problems

[rconaway]

I deployed 4 nanostations around a baseball field about 70 feet in the air with the NS2's in AP WDS mode using external antennas. All four radios talk to each other with no problem and link speeds are 54Mbps. I can ping through to any of them and have no problem opening up the web interface. However, what I can't do is connect to them consistently. It doesn't matter where I am. One try out of 10 I can connect. If I reboot them I might be able to connect for a few moments. I have tried static IP's and a router on the back side as DHCP. However, my real problem is getting the connection part. I changed the distance from .2 miles to 1 mile and it seemed to help after rebooting. Then I lost it again.

This is frustrating because I took down a Tropos network that was working and I'm sitting with egg on my face right now. I'm kind of out of ideas with settings.

[UBNT-Mike.Ford]

Hello Rconway,

If all 4 units are in AP/WDS mode, and can see each other on the same SSID, then I can almost guarantee that you are looping your network preventing the data from bieng passed.

thank you,

Mike

[rconaway]

Sorry Mike, I said this incorrectly. Only 1 AP has the three MAC addresses in the WDS links. The other three AP's have only 1 MAC Address, the base unit. It's basically a PTMP setup. I was just describing the physical setup.

[rconaway]

New update: The main AP is no longer talking to the other units. It can see them in the site survey, it just can't connect to them. This happened over a 24 hour period. We are going to reboot all of them to see if they link up again.

[rconaway]

I am doing a similiar situation. Here is what we are finding.

1) The WDS came up and with one AP wired to a server. From the server, we can ping everything.

2) We had great difficulty connecting to any AP from a laptop. Once we were connected, everything seemed to work fine for a few minutes and we could ping everything.

3) A few minutes later, we could not longer connect although the server could still ping every radio meaning the WDS backhaul was working.

4) If a WDS goes down, sometimes you have to reboot both AP's to come back up. Once they disconnect, sometimes you can usually connect to them with the laptop and reboot them. However, it's intermittent and one of them we had to power cycle.

5) After 24 ours, we lost 2 of the 3 WDS links. Rebooting fix that but the radios are about 500' apart with 15dBi antennas with 5' LMR-195 cables. We don't have a signal problem.

Based on the current deployment, we have come to the following conclusions so far.

1) Using several radios in a WDS mode is not working they way we thought they would. The radios aren't capable of recovering a WDS link without reboots. This would not be a problem with Watchdog but we are just now testing with that.

2) The difficulty we had connecting with a laptop at distances of a couple hundred feet or more was interesting. We just pulled out Tropos radios and they seemed to work fine at those distances so we are confident the three laptops used are good. We have 15dBi omni's on the radios and in this test, we aren't seeing a good consistency of connection. It's possible there is a bad radio in this group but I haven't seen a bad unit out of 40 units so far in station mode.

3) Apparently we need help get this configuration running. I really want this to work for our applications as we would like to start deploying this in two other cities but things aren't looking good.

We have about a week to get this figured out or we have to get another product in there. Any help would be greatly appreciated.

[UBNT-Zy]

Hello,

please upgrade the units to the latest FW, available there:
http://www.ubnt.com/downloads/XS-fw/v3.1/

The similar AP-WDS to AP-WDS test network is running in our lab with the latest FW for hours.
4 stations connected to different AP-WDS can ping all the other. I'll leave this setup with continuous pings for few days.

Please check if all the configuration requirements described in the following topic are met:
http://ubnt.com/forum/viewtopic.php?p=8259

[rconaway]

The latest firmware lets them ping each other. However, it still doesn't pass DHCP from a DHCP server, Microsoft or otherwise. In addition, the IBM laptops still can't connect with static IP addresses. I can provide a direct RAS login into the server attached to the primary nanostation if you want to see it. We have dude running and I can have a tech on site while you are in the AP's.

[rconaway]

Ignore that last message. We tested last night with firmware we got yesterday, NS2.ar2316.v3.1-rc.3345.080620.0045.bin. We will restest with the firmware that was uploaded today. As I have mentioned before, we can get you a direct login into a Windows server that is physically attached to the primary NS2 if that will help. We can then place a tech on-site with the problem laptops. The windows server is the DHCP server that won't pass through.

[bjohnston]

Hello everyone, I am one of the Techs working on rconaways network,
Here is the physical layout that is existing. TheNS2's are on the 4 corners of a square like so:

AP1---------------------------AP2








AP3---------------------------AP4


AP1 is Ethernet wired to the lan where DHCP server and Internet router is.
AP1 is in AP WDS mode with AP2,AP3,AP4 all WDS back to AP1

so all 4 units are in AP/WDS mode in order to provide full coverage of the property.

AP1 has the MAC of AP2,AP3,AP4 in the WDS setup.
AP2 has the MAC of AP1
AP3 has the MAC of AP1
AP4 has the MAC of AP1

With the 3.1RC firmware all WDS links were up and stable, but I could not connect with an IBM laptop and internal card. If i used a prism or atheros based chipset i can connect if i use a static IP. but not if I use DHCP.

I upgraded to the 3.1 Full release just now and now the wds links will come up for a couple of seconds then go down and stay down. I can get them to come back up again for a few seconds if I reboot AP1 (main) but they will lose connection after a couple seconds and stay down.

Thanks
Brian

[UBNT-Zy]

Hello Brian,

Thanks for clarifying. So you are using star shaped Point-to-Multipoint topology, while 3 AP-repeaters are connected to one AP.

Can you please specify the IBM laptop model and the exact chipset version of the internal card, including operating system version and internal wireless card driver version.

DHCP client functionality should not depend on the wireless chipset vendor - it seems this is connectivity issue. There may be DHCP configuration issues in STA-Bridge, but not in WDS as WDS is fully transparent mode.
Make sure you have disabled the option "Enable Extra Reporting" in all the AP units.

Can you please connect the laptop to one of the repeater access points (AP2,AP3,AP4) ethernet port and check if it can retrieve DHCP settings on the LAN port.

The possible root cause of the issue while WDS "go down and stay down" can be related to the temporal or permanent network loop - i.e. a laptop is connecting to any of the repeaters using the wireless interface, while it is connected with ethernet to the same network.
Can it be that the laptop is continuously switching from one AP to the another one?
Let's eliminate these possibility with the folowing test:
- Enable the "MAC ACL" on all the 4 AP, while setting the Policy to "Deny".
- Check if the inter-AP connections are stable (you can connect laptop to ethernet port of each AP to check this, or telnet/ping each AP from the DHCP server)
- If everything works as expected, Disable the "MAC ACL" on the first AP, connect the wireless station to it and repeat the test.
- If everything works as expected, Disable the "MAC ACL" on the second AP and try to replicate the issue.

[rconaway]

The IBM's T42 uses an Atheros chipset. Can't tell you which one but the driver version is 4.1.2.156

I'm not sure I agree with your DHCP assessment of WDS. We can demonstrate wireless network cards that work with static IP addresses but never completes a connection with DHCP. They just time out. If we hard wire into the network and turn off the wireless card, it works fince which shows the DHCP server was working. We tried 2 different DHCP software packages, one was window 20003 server. However, these same wireless network cards work fine everywhere else.

We can't connect directly to the current AP's without bringing a lift on site which will cost $9000 (The radios are 80' in the air). We will bring a 5th AP on-site and test with that at ground level and add it to the WDS lists. If further testing shows promise, we will bring the lift.

No laptop was physically connect to the netork to creat loop.


Hello Brian,

Thanks for clarifying. So you are using star shaped Point-to-Multipoint topology, while 3 AP-repeaters are connected to one AP.

Can you please specify the IBM laptop model and the exact chipset version of the internal card, including operating system version and internal wireless card driver version.

DHCP client functionality should not depend on the wireless chipset vendor - it seems this is connectivity issue. There may be DHCP configuration issues in STA-Bridge, but not in WDS as WDS is fully transparent mode.
Make sure you have disabled the option "Enable Extra Reporting" in all the AP units.

Can you please connect the laptop to one of the repeater access points (AP2,AP3,AP4) ethernet port and check if it can retrieve DHCP settings on the LAN port.

The possible root cause of the issue while WDS "go down and stay down" can be related to the temporal or permanent network loop - i.e. a laptop is connecting to any of the repeaters using the wireless interface, while it is connected with ethernet to the same network.
Can it be that the laptop is continuously switching from one AP to the another one?
Let's eliminate these possibility with the folowing test:
- Enable the "MAC ACL" on all the 4 AP, while setting the Policy to "Deny".
- Check if the inter-AP connections are stable (you can connect laptop to ethernet port of each AP to check this, or telnet/ping each AP from the DHCP server)
- If everything works as expected, Disable the "MAC ACL" on the first AP, connect the wireless station to it and repeat the test.
- If everything works as expected, Disable the "MAC ACL" on the second AP and try to replicate the issue.

[bjohnston]

Hello Brian,

Thanks for clarifying. So you are using star shaped Point-to-Multipoint topology, while 3 AP-repeaters are connected to one AP.

Can you please specify the IBM laptop model and the exact chipset version of the internal card, including operating system version and internal wireless card driver version.

DHCP client functionality should not depend on the wireless chipset vendor - it seems this is connectivity issue. There may be DHCP configuration issues in STA-Bridge, but not in WDS as WDS is fully transparent mode.
Make sure you have disabled the option "Enable Extra Reporting" in all the AP units.

Can you please connect the laptop to one of the repeater access points (AP2,AP3,AP4) ethernet port and check if it can retrieve DHCP settings on the LAN port.

The possible root cause of the issue while WDS "go down and stay down" can be related to the temporal or permanent network loop - i.e. a laptop is connecting to any of the repeaters using the wireless interface, while it is connected with ethernet to the same network.
Can it be that the laptop is continuously switching from one AP to the another one?
Let's eliminate these possibility with the folowing test:
- Enable the "MAC ACL" on all the 4 AP, while setting the Policy to "Deny".
- Check if the inter-AP connections are stable (you can connect laptop to ethernet port of each AP to check this, or telnet/ping each AP from the DHCP server)
- If everything works as expected, Disable the "MAC ACL" on the first AP, connect the wireless station to it and repeat the test.
- If everything works as expected, Disable the "MAC ACL" on the second AP and try to replicate the issue.

Here is your answers:

OK, I did a packet capture from tonight, It is about 28000 packets and is only a 20 second sample. yep, 28,000 packets in 20 seconds. I followed the instructions on the forum post and nothing made any difference.

Here are the answers to your questions and suggestions in your post.


Can you please specify the IBM laptop model and the exact chipset version of the internal card, including operating system version and internal wireless card driver version.

It is an Intel PRO/Wireless 2200BG running driver Version 9.0.4.37

DHCP client functionality should not depend on the wireless chipset vendor - it seems this is connectivity issue. There may be DHCP configuration issues in STA-Bridge, but not in WDS as WDS is fully transparent mode.
Make sure you have disabled the option "Enable Extra Reporting" in all the AP units.

"Enable Extra Reporting" is disabled in all units

Can you please connect the laptop to one of the repeater access points (AP2,AP3,AP4) ethernet port and check if it can retrieve DHCP settings on the LAN port.

Cant connect to any of the endpoint radios as they are 80' up a Pole, but I can connect to Ethernet on the inside of the wired LAN and get DHCP fine.

The possible root cause of the issue while WDS "go down and stay down" can be related to the temporal or permanent network loop - i.e. a laptop is connecting to any of the repeaters using the wireless interface, while it is connected with ethernet to the same network.
Can it be that the laptop is continuously switching from one AP to the another one?

Nope, stays connected to a single AP until I get out of range then will switch. I verified this with a packed sniffer.

Let's eliminate these possibility with the folowing test:
- Enable the "MAC ACL" on all the 4 AP, while setting the Policy to "Deny".
- Check if the inter-AP connections are stable (you can connect laptop to ethernet port of each AP to check this, or telnet/ping each AP from the DHCP server)

Nope, cant ping anything from Server,


- If everything works as expected, Disable the "MAC ACL" on the first AP, connect the wireless station to it and repeat the test.
- If everything works as expected, Disable the "MAC ACL" on the second AP and try to replicate the issue.



I believe that the underlying problem is causing extreme amounts of traffic and that is what is causing the WDS links to become unstable. I can ping the remote units from the ping tool in the main AP that is connected to the LAN, but not from anything on the wired LAN itself.

Something else that I tried is enabling "Spanning Tree Protocol" on all radios and this made no difference either.

I e-mailed you a packet capture that is from a 20 second window so you can see the amount of traffic that is being passed. There are a TON of Control/ACK and CTS packets.

Hope this all makes sense as I'm tired and its been a long day.

Brian

[UBNT-Zy]

Hello Brian,

thanks for the answers and really good report.

there is the new FW version available in our site:
http://www.ubnt.com/downloads/XS-fw/v3.1.1/

It has a WDS functionality improvement which should eliminate network looping between several AP-WDS peers if STP is enabled.

Z.

[bjohnston]

OK, that fixed the massive traffic issue which allows the WDS links to be stable now. and I can ping all of the APs and network nodes from anywhere on the network.

Here are the remaining issues.

I can not connect with a client using DHCP no matter what brand of laptop or chipset. But if i assign a static IP address to the client i can connect and have full network functionality, also no matter what laptop or chipset.

I will be forwarding 2 more packet captures, one from connection with DHCP and one with connection with Static IP.

Brian

[sminiea]

This problem seems to affect me also with same network config.

(1) Bullet2 connected directly to internet via lan. AP/WDS setup w/ macs of 3 other bullets. Error reporting off. Auto wds off.

Pinging each bullet thru lan is no problem. WHen I connect wirelessly - I experience the same probs as above.

Changing the follwong setting in laptop from previous post - see immediately below .........

"... i im using acer machine with Atheros AR5006X Wireless card. i change to a intel pro wireless card and i receiving packages and the internet is working. so i install again the atheros and change some setings in the advaced tap.


MFP - ENABLE
POWER SAVE MODE - DISABLE
SCAN VALID INTERVAL - 120"



............ fixes these problems. I need a fix that won't require me changing these settings in each laptop that will ever use this. I imagine there is a fix in the AP/bullet settings.

Please help!