MGMT-CONN saturation

[Fernando Bello]

Hello All,

We have 2 ACE20 Modules in 2 different Cat 6500. We have several contexts, some are active on one Cat6k and some are active on the other, just to loadbalance the traffic through the Cat6k switches. We found that on only one context that is in standby state the mgmt-connections are left out of resources and are being denied.

ACE2/C1# sh resource usage

                                                     Allocation

        Resource         Current       Peak        Min        Max       Denied

-------------------------------------------------------------------------------

Context: C1

  conc-connections           2428       6516    1600000          0          0

  mgmt-connections          19998      20000      20000          0    2895771

We also noticed that several probes from the standby C1 context are failing.

On the C1 active context, everything is working fine.

The version is:

Software

  loader:    Version 12.2[121]

  system:    Version A2(3.0) [build 3.0(0)A2(3.0)]

  system image file: [LCP] disk0:c6ace-t1k9-mz.A2_3_0.bin

  installed license: ACE-08G-LIC

Hardware

  Cisco ACE (slot: 2)

  cpu info:

    number of cpu(s): 2

    cpu type: SiByte

    cpu: 0, model: SiByte SB1 V0.2, speed: 700 MHz

    cpu: 1, model: SiByte SB1 V0.2, speed: 700 MHz

  memory info:

    total: 954976 kB, free: 217880 kB

    shared: 0 kB, buffers: 5920 kB, cached 0 kB

  cf info:

    filesystem: /dev/cf

    total: 1014624 kB, used: 391760 kB, available: 622864 kB

last boot reason:  reload command by csperanza

configuration register:  0x1

We have 5 contexts + the Admin.

First I would like to clear the 19998 conn in the standby context. We did a clear conn all and nothing happened.

Is it possible to check what are the connections that might be causing problems?

We did a show conn but it shows the conc-connections, not the Mgmt connections.

Thanks in advance

[Kanwaljeet Singh]

Hi Fernando,

The connections should show up in show connections i.e when you telnet to ACE interface, you should see an entry in show conn table.

You can try to shut  and no shut the management interface and see if it clears the connections. You should also see that  denied counter should stop increasing after that.

Also, the probe failures are probably due to resource crunch. Can you see what is failure reason for probe if you do show probe detail?

You can also try and increase the resource allocation for management conn.

Let me know how it goes.

Regards,

Kanwal