Losing access to network

[Craig-Lyndes]

Confirmed that problem persists.

Software version: 5.5.3.0-01-10326

MSM 760 Controller

MSM 410 Access Point

Set to B,G,N experience lockups where clients see SSID and show signal strength but can not reach network or Internet.  Local servers, DHCP, DNS and Default router unreachable.  Access Point is still visible from wired ethernet network.  Responds to Reset command from MSM760.  Reset or powercycling the access point solves the problem.

This has happened 3 times.  I am resetting the MSM 410 to G only.  Am unable to do this on the new MSM 430s!

5.5.3 seems to not have solved the problem.

Craig Lyndes

St Albans City School

[ISoliman]

Please collect the filtered and unfiltered logs from the controller and upload them to the post.

Is your issue affecting all type of clients or only specific laptops/devices?

Can you please upload the config also? and make sure under the VSC disable/uncheck the "wireless security filter" option as this will restrict the traffic to the controller and will reset/stop the connection after a while if the connection is idle.

[Craig-Lyndes]

I can't give you the log because the time that the access point quit has already rolled off.  I have reset the access point to B,G,N and will wait for it to happen again and then will capture the log.

When it happens it effects all clients connected to that access point.  In many months of observing this I have noticed that it most often happens with there is a Mac (airport chip set) added to the large number of Intel, etc chip sets that are most common in our environment - we are primarily Windows with only a few Macs. 

I do not have the "wireless security filter" set.  Open, unencrypted, wireless network.  One SSID, no VLANs. 

Craig Lyndes

St Albans City School

[payne324]

Hi we are seeing the exact same symptoms and can make a similar correlation to the Mac "airport chip set" as you mentioned. We have 40 MSM710 controllers and 8 MSM760 controllers running firmware version 5.4.1 on the majority of controllers/aps and 5.5.3 on a few and see it everywhere. We have over 400 MSM410 access points. Currently we can only run at G only (2.4Ghz) or 5Ghz N. The moment we switch to 2.4Ghz G and N on our AP's we see the symptoms listed "lockups where clients see SSID and show signal strength but can not reach network or Internet.  Local servers, DHCP, DNS and Default router unreachable.  Access Point is still visible from wired ethernet network.  Responds to Reset command from MSM760." We cannot switch all APs over to 5Ghz N as we do support a VSC for Public Guest access so teachers and students can utilize their own wifi devices in the schools. This has become a big pain point. The lockups happen on both an access controlled VSC and a non access controlled VSC. The client never becomes disassociated with its AP or the controller and the AP's uptime and Controllers are good.

Hopefully there will be a firmware fix from HP soon.

Steve Payne

Hastings and Prince Edward District School Board

[Craig-Lyndes]

ISoliman

Here are the log files and the configuration file.  The access point is a MSM410.  The controller is a MSM760.  The access point is named Server Room.  You will see that at the time I took the configuration I had reset it back to 802.11G only.  The crash was about 5 minutes previous to the time the log was taken.  Look for the access point starting up.

Craig Lyndes

St Albans City School

OK, so this forum doesn't want to make uploading easy.... 

[ISoliman]

Without going further after checking the logs I found the below:

BID check failure

this is caused by an issue with the internet port, please contact the support so that they can fix this issue for you, the link below is officially from HP regarding this issue

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?lang=en&cc=us&objectID=c03054362

After this is fixed check if the issue is still showing.

[ISoliman]

Also to add to this, from the latest logs I can see that "N" is still on, check below logs:

AutoChannelSelection: ifname=wvlan0, selecting best initial channel (1,-89dBm)
Nov 30 12:47:37 debug    kernel       SG9202C0KX 11n protection mode has been enabled due to the presence legacy access points
Nov 30 12:47:37 debug    kernel       SG9202C0KX 11n protection mode has been enabled due to the presence legacy clients

See the last regarding the protection mode, this means N is not disabled.

The last thing, from below logs (as you mentioned the issue showed approx. 5 mins before resetting the AP:

Nov 30 12:29:33 debug    kernel       SG9202C0KX
Nov 30 12:29:33 info    kernel       SG9202C0KX wvlan0: decreasing transmit power to 10
Nov 30 12:32:23 debug    monitord     SG9202C0KX Stopping: <mgetdate>
Nov 30 12:32:24 debug    mgetdate     SG9202C0KX SNTP time synchronization using server secure_msc
Nov 30 12:32:42 debug    openvpn      SG9202C0KX TLS: soft reset sec=0 bytes=10245766/0 pkts=60484/0

you can see that before the reset the AP decreased the power to 10, this might cause the clients away from the AP to loose connection and not able to connect, I will not recommend anything now since i didn't check the config yet but you can try and disable the auto power control option (after solving hte BID issues and calling the support) then see if it works fine or not.

[Craig-Lyndes]

ISoliman,

The reason it is upset about DHCP on the Internet port is there is nothing connected to the Internet port.  The controller is not routing.  It is only handing out configuration to the access points and gathering statistics.  The wireless network would run without the controller as long as the access points were not power cycled.

The only connection to the access point is the LAN (Port 2).

Craig Lyndes

St Albans City School

[ISoliman]

Yes I noticed that you are using the controller for management only, but this error is critical so you better fix it, call the support they will ask for specific thing and provide you with the fix, also I noticed you are using firmware version 5.5.3.0, are you entitled for it ? if not then I would recommend using any other version for 2 reasons, 1st because you are not entitled for it (if), 2nd is because this branch is not that stable yet as it was made to support the new 460 466 & 430 APs, also check the config and make sure to disable the auto power control thing, one more thing is to do a site survey as the controller might be decreasing the power of the AP to the minimum to avoid an interference.

[Craig-Lyndes]

ISoliman,

You are correct.  When the access point was rebooted it was still in b,g,n mode.  I reconfigured it after the reboot, so what you see in the logs would reflect the b,g,n configuration.  After I captured the logs I reconfigured the access point, then I exported the configuration.  Sorry to muddy things up.

The vairable transmit power is relatively new.  I have used it both set at 100% and at auto power.  My access points are close together, so I hoped that by using the auto power I could decrease the RX FCS errors, which are very high.  This is a separate problem from the clients getting kicked off. (maybe?)

The event that I tried to capture effected every device connected to the access point.  My notebook was within 5' of the access point with nothing but air between.  My notebook was unable to reach DNS, Default Gateway, DHCP server or other servers.  The access point was however visible to the controller and showed no issues (which is why the logs don't seem to be of much use, at least for this problem).

Please keep looking though, there has to be a reason.  And if you see anything at all bring it to my attention.  And if my "rationalizations" seem flawed please point the flaws out to me.  I have spent a lot of time trying to resolve this problem, and it seems to be unsolvable.  Having better minds than mine look at it is very appreciated.

I'm very suspicious of the routing inside the access point.  This is because the radio side seems to be functioning and the ethernet side also functions.  This doesn't support the solution, only use 802.11G, what does that have to do with routing?

Craig Lyndes

St Albans City School