CSS - keepalive frequency, retryperiod and maxfailure

[sidewinder]

Hello,

i have a question about the keepalive and Service state.

There is a keepalive frequency which is the time between normal keepalives are sent, when the service is alive.

So my question goes specialy to the retry-period.

If the keepalive response to my keepalive was not correct, then my service will go in a dying state till the number of max failure are reached. The the service state will be down, if any of the keepalives where wrong.

But when exactly the keepalive retryperiod will be uses. Is this applied when the Service is in 'down' mode ore in 'dying' mode.

Example:

keepalive-frequency 20

keepalive-maxfailure

keepalive-retryperiod 60

The time, till down is detected is now be calculated as:

(frequency * maxfail) + (frequency-1)

or

is this wrong?

Best Regards

[stevehall]

The frequency is how often we send a keepalive to a service that is not down. Dying state services still get keepalives at the normal frequency.

The retry period is how often we send a keepalive to a service that is down... (If it is down, we may not want to check it every 5 seconds, but every minute instead)

-Steve

[Sbutzek]

Thanks,

that was, what i assumed.

Best Regards

[Sbutzek]

Hello,

i have tried this in my lab. It tested a critical service with different retry-periods.

And this does not work, how i assumed this.

If i configure a maxfail of 2, a frequency of 10 and a retryperiond of 50, then it takes about 60 seconds till my critical service will get down.

If i change the retryperiod to 5 seconds, then the service is much faster down and my vrrp process is changing.

That is not how the retryperiod was explained. It seems to be, that the retryperiod is used when the service gets to state dying.

Best Regards

Sven Butzek

[Gilles Dufour]

Sven

the explanation given to you by Steven is correct.

The way you verify it seems to be incorrect.

Use a sniffer trace to check the frequency.

How do you brindge the service down ?

Looks like you disconnect CSS cables. Try instead to disconnect the server so you can use a sniffer to see the CSS sending the probes.

Regards,

Gilles.

[Sbutzek]

Hello Gilles,

okay i will try to explain what i did.

Having 2 CSS, and 2 Cat 3550.

Each CSS is connected to one Cat 3550, the Cat 3550 are connected together:

Cat 3550 --- CSS1

|

|

Cat 3550 --- CSS2

So if i shut down the interface vlan on the cat3550 which i check over the keepalive, then i see the timers of the retryperiod. The physical links are always up.

Best regards

Sven Butzek

[Gilles Dufour]

Sven,

my mistake and sincere apologies.

The retryperiod is whenever a service is not alive.

This include the dying state.

This is documented at :

http://www.cisco.com/en/US/products/hw/contnetw/ps789/products_command_reference_chapter09186a00800e4519.html#1028271

When a service has failed to respond to a given keepalive message (the service has transitioned to the dying state), the retry period specifies how frequently the CSS tests the service to see if it is functional.

I did some test in the lab and I could confirm this.

Regards,

Gilles.