CSS Balancing Laod too much connection

[abelleli71]

Hi

Need a feddback about this issue

I have two CSS 11503 with load balancing active  on # 03

This machine register an high charge of connection and server Front end are very slow to log in  ( the ICMP reply works )

Did someone can see something wrong ? ( logs are clear and without any error )

THANKS

Alberto

rrmr003# show load
Global load information:

    Reporting:Enabled

    Calculation method:Relative
                       Step Size:Dynamic Configured:10  Actual:10

    Threshold:254    Ageout-Timer:60
    Teardown-timer: Configured:20  Actual:2

Service load information:
                                  Average      Average     Peak Average
       Service Name             Load Number  ResponseTime  Response Time
-----------------------------------------------------------------------
RRMH004                         2            0             2110405
RRMH005                         2            619427        1893706

rrmr004# show load
Global load information:

    Reporting:Enabled

    Calculation method:Relative
                       Step Size:Dynamic Configured:10  Actual:10

    Threshold:254    Ageout-Timer:60
    Teardown-timer: Configured:20  Actual:20

Service load information:
                                  Average      Average     Peak Average
       Service Name             Load Number  ResponseTime  Response Time
-----------------------------------------------------------------------
RRMH004                         2            0             0
RRMH005                         2            0             0

[yushimaz]

In general, high value of response time occurs due to slowness of server response.

So, I suggest you check the load of server application.

In the meantime, I'll check CSS.

What kind of probe do you configure on RRMH005?

Please let me know your keepalive configuration?

And also please get the following log?

CSS11503# llama

CSS11503(debug)# flow statistics

CSS11503(debug)# show system-resources

CSS11503(debug)# icp probe service RRMH005 [url by quoted text]

Regards,

Yuji

[abelleli71]

Hi Yuli

Thanks for the answer

hi the the configuration of the keepalive

service RRMH005
  ip address 10.192.149.25
  port 389
  protocol tcp
  keepalive type tcp
  keepalive port 389
  redundant-index 2
  active

Find also in attach some other output

Alberto

[yushimaz · Accepted solution]

Hi Alberto

I checked your configuration and looked fine.

So, if your problem is related to the high value of reponse time,

it maybe a server issue.

Regarding low performance issue of CSS, it may be related to

the flow idle timeout value. I've handled perfomance issue as

TAC SR and some of them were related to the idle timeout.

Since idle timeout of tcp is 16 seconds by default, flow will be

deleted from active flow list (show flows) if flow is idle more than

16 seconds.

If you want to change this idle timer, please configure as below.

CSS11503-A(config-owner-content[test-test])# flow-timeout-multiplier 100

BTW, your CSSes run about 800 days. I suggest you upgrade WebNS

or reboot CSSes since there are some bugs regarding 828 days issue

as follows:

CSCek37489 : CSS: Backup CSS sends invalid VRRP packets after long running

CSCsy38035 : CSS: After long run, Master of CSS detect Dup vrrp packet

Regards,

Yuji

[abelleli71]

Hi Yuji

Thanks soo much for your suggestion

I will notify to client !