Network Management

Hello,

I have a wierd problem with one of my airwave server. At my main campus, I have a airwave server just for that campus and then I have a master console for getting all data from 3 airwave servers (1 in the main campus and 2 at my remote campus). Last year, we had about 5000 concurrent users at the start of the school year. This year however, we have seen a massive explosion of concurrent users on our network (maxing out at aprox. 14k users).

My airwave server is having problems with snmp and ping in fact. The symtoms are my snmp and pings will fail multiple times, but will suddenly work 1 time in a persiod of 15 mins. I have checked all my settings (as far as I can think off).

In the cli, when i perform a snmp walk, this is the result I get:

ping 10.29.100.26
connect: Resource temporarily unavailable

s2w 10.29.100.26 mysnmppwd
snmpbulkwalk: Failure in sendto (Resource temporarily unavailable)

It seems to be a resource issue and I have not being able to figure it out. My master console can ping with out a problem to the same IP address. So i am wondering if anyone has encoutered this before?

 My Airwave spec:

My airwave server is monitoring 1900 devices and aprox 140000 concurrent users at peak.

Regards,

Edward

The issue is directly related to size of the SNMP bulk get.  When the table is super large, it takes longer for the controller to build up (be careful to monitor the controller utilization).  Depending on which controller type you have, this table can grow to a point that AMP will continue to have such problems.  The direction should be to move from SNMP to AMON for client data (AMP Setup -> set Prefer AMON vs SNMP = yes).  At the same time, the controller needs to have the correct profile to send AMON messages to AMP (req at least controller AOS FW 6.2, but latest release is advised).

example from a controller cli:

# show run | include mgmt

mgmt-server type amp primary-server x.x.x.x profile default-amp

# show mgmt-server profile default-amp

Mgmt Config profile "default-amp"
-----------------------------------------------
Parameter Value
--------- -----
Stats Enabled
Tag Enabled
Sessions Enabled
Monitored Info Enabled
Misc Enabled
Location Enabled
UCC Monitoring Disabled
AirGroup Info Disabled

The use of AMON reduces load on controller from SNMP requests.  It also provides data to AMP on a regular basis.  If you're not on the latest 8.0 release, then auditing needs to be enabled from the AMP side.  Correct controller credentials for ssh/enable need to be in the controller manage page as well - for use of requesting data for missed message types.

I have already setup my controllers to use AMON. The only change I had to make in the default profile for AMON was the following:

monitored-info-enable

monitored-stats-enable

One question, you have indicated in Airwave to "set Prefer AMON vs SNMP = yes", but I have also seen in another post that this setting should be left alone, i am reading it wrong?

http://community.arubanetworks.com/t5/Unified-Wired-Wireless-Access/AMP-8-0-4-mysteries/td-p/203309

Regards,

Edward

I'll have to take a closer look at that other thread.

But for your situation, it sounds like the SNMP load is bogging down the controller.  So in order to reduce that, setting Prefer AMON vs SNMP to yes will cause the AMP to no longer poll controllers over SNMP for the client table.

When AMON is enabled on the controller side, AMP doesn't automatically start processing.  The first enable AMON setting will allow for processing of RF Health, client speed, and client goodput.  The Prefer AMON vs SNMP is currently for moving from polling controller for client data to parsing the data from the AMON feed.  There's a few open issues, but aside from the known open cases, we've heard good things from customers we've worked closely with.

If you're seeing discrepancies after setting the Prefer AMON flag, open a support case so that we can address the issue.