Lessons learned on Cisco Catalyst 4500E

At work we switched from a Cisco Catalyst 3560G to a Catalyst 4500E a few weeks ago.

Remember to issue the spanning-tree extend system-id command prior to enabling VTP version 3.

During the test period, the Icinga instances logged packet losses of varying degree around the clock when doing ICMPv6 pings.

** PROBLEM Service Alert: [hostname]/PING6 is WARNING **

***** Icinga *****

Notification Type: PROBLEM

Service: PING6
Host: [hostname]
Address: [address]
State: WARNING

Date/Time: [withheld]

Additional Info:

PING WARNING - Packet loss = 20%, RTA = 65.98 ms
** PROBLEM Service Alert: [hostname]/PING6 is CRITICAL **

***** Icinga *****

Notification Type: PROBLEM

Service: PING6
Host: [hostname]
Address: [address]
State: CRITICAL

Date/Time: [withheld]

Additional Info:

PING CRITICAL - Packet loss = 60%, RTA = 0.72 ms
** RECOVERY Service Alert: [hostname]/PING6 is OK **

***** Icinga *****

Notification Type: RECOVERY

Service: PING6
Host: [hostname]
Address: [address]
State: OK

Date/Time: [withheld]

Additional Info:

PING OK - Packet loss = 0%, RTA = 0.77 ms

The 3560G was far more happier than the 4500E with a DHCP(v4) lease time of 5 minutes for one of the VLANs. After nearly a week in operation, the 4500E suddenly stopped acting as the DHCP(v4) relay for that VLAN. Other VLANs were unaffected.

Increasing the lease time to 10 minutes and reloading the switch made the 4500E much happier. No more packet losses has been logged so far.

Note to self, remember to check the configuration register prior to reloading the switch with high hopes of loading a new version of IOS XE. The value 0x2102 is so much better than the default of 0x2101.