3750 Stack- Master "down" recovery time

[Map23]

Hello forum, 

I did many test to understand which is the real recovery time of Catalyst 3750 stack when Master goes down.

From test I get different results (from 10 seconds to 2 minute recovery time), but I don't understand why I obtain these different time, and which are conditions for best performance. 

Can you help me to understand (or suggest test) which are components that influence that recovery time ?(memory status, active protocols etc) 

Thank you

[Reza Sharifi]

Hi,

Can you explain how you are doing the test?

What components and tools are you using?

[Map23]

Googd morning Reza, 

I'm sorry, my "did many test" was a very generic phrase :) 

I did all test with multicast configurations. When I use only L2 multicast confs results are always about 10 seconds but when test include L3 multicast confs I have different results. As from theory L2 continuity is right, because on that document: 

http://www.cisco.com/c/en/us/products/collateral/switches/catalyst-3750-series-switches/prod_white_paper09186a00801b096a.html

I found: 

When one master switch becomes inactive and while a new master is elected, the stack continues to function. Layer 2 connectivity continues unaffected

Note that for multicast L3 configuration I mean only 2 vlan (1 source and 1 receiver) with PIM sparse mode and IP multicast e and IP routing enabled. Recovery time is bounded to L3 recovery (Interfaces back to up state). But this recovery doesn't happens  always at the same time, often after 10 second but other times after 2 minute. 

I wish that tests are now clear, 

Thank you for your help