When interfaces on a Cisco 3750 are configured as part of a port-group using LACP (passive) what are the timers or delays associated with putting one of the ports into "suspend"? What is required to get a port out of "suspend"?
Here is the scenario we saw prompting me to ask this question. First, here is the hardware involved and how they are connected. ServerA is a Solaris server running Solaris 10. ServerA has three NICs in use. NIC1 is connected to Switch1. Switch1 (Cisco 3750) is on a network with access to the Internet. NIC2 and NIC3 are connected to Switch2 (also Cisco 3750) using LACP mode active on the server side and passive on switch side. Switch2 is on a private network used only for iSCSI SAN connections.
Second, here is the problem that started the unfortunate chain of events. A layer 2 loop was accidently created on Switch1 which caused ServerA to reboot.
Third, here is the effects of the reboot. When ServerA came back up, NIC2 and NIC3 both reported that the LACP configuration was in place but that only NIC3 was connected but could not send any traffic. On Switch2 the interface for NIC2 was "notconnected" and NIC3 was "suspended".
Fourth, finally, here is what we did to get the LACP connections between ServerA and Switch2 back up and working. We removed all of the configurations for LACP on both ServerA and Switch 2 and could demonstrate one at a time that each interface worked (was able to send and recieve traffic). But we could not re-establish LACP between those interfaces. What did allow us to get LACP working again was to disconnect NIC2 and NIC3 from Switch2, reconfigure new interfaces on Switch2 to use LACP, configure NIC2 and NIC3 for LACP again and reconnect NIC2 and NIC3 to the new interfaces on Switch2. This time LACP worked properly, even when rebooting ServerA. So, in an effort to understand the problem, we reconfigured the original interfaces on Switch2 for LACP and moved back to them. This time LACP worked without a problem and no ports were "suspended".
We are trying to understand why LACP did not work after the original reboot of ServerA but was able to work on the same interfaces when the same configuration was applied less than an hour later. When interfaces on a Cisco 3750 are configured as part of a port-group using LACP (passive) what are the timers or delays associated with putting one of the ports into "suspend"? What is required to get a port out of "suspend"?
Any help you can provide in understanding the behavior of the Cisco 3750 switch in what is descrived above is greatly appreciated.