We have 2 x 6513's both with one FWSM and ACE module in each (active / standby for both).
We have been having some issues with the cats and as such we've been informed to upgrade to 12.2(33)SXH2a, specifically to correct the following issue:
Due to the presence of the EtherChannel across DFC cards and HSRP being balanced active across the core switches running SXH1 (the original IOS), signs were pointing towards CSCso29141 (DFC installs drop index for MAC-address after -ve MN). The mac-aging-time was at the default 300 rather than the recommended 480 when MAC synchronization is enabled. Following the workaround for CSCso29141, we've statically configured 'mac-address sync', increased the aging time with 'mac-address aging-time 480', and decreased the ARP timeout for all active VLANs to 300 secs with '(config-if)# arp timeout 300'.
Now we've upgraded the switches and also applied the workaround.
However, when rebooting either 6513 now, it seems the FWSM can not be in the switch (FWSM is in slot 1, ACE in slot 2) as these are being powered up too soon . These modules need to be removed before rebooting these switches at this stage. I believe this is because the 10 gig cards take much longer to boot then the FWSM which allows this module to start talking to the network before the switch is connected to the network - hence causing the FWSMs to think they are both primary (eg. they both are 'up' before their failover vlan becomes up).
Does anyone know how we should get around this issue?
Specifically, we would like a situation where we could bring up a 6513 without having to worry about if the underlying vlans are up or not prior to the FWSM's booting. We've logged a call with the TAC but was wondering if anyone has seen these issues before...