Symptom

On a C9600X or C9500X in stackwise-virtual, after a reload or power-cycle, ports could show as down or err-disabled. This behavior is specific to dual-rate optics such as SFP-10/25G-LR-S, on linecards or switches that have port-groups (this includes C9600-LC-48YL, and C9600-LC-40YL4CD, and C9500X-60L4D). Ports in a port-group will detect a config-mismatch between other port in the port-group, resulting in one or both interfaces in the port-group being down or errdisable. Logs that may be seen (if errdisable detect cause sfp-config-mismatch is not configured, will not see this error message): %PM-4-ERR_DISABLE: sfp-config-mismatch error detected on Fif1/1/0/41, putting Fif1/1/0/41 in err-disable state

Conditions

C9600X or C9500X in stackwise-virtual running on software versions 17.10.1 and above. Hardware: • C9600-LC-48YL • C9600-LC-40YL4CD • C9500X-60L4D • Dual-rate optics such as SFP-10/25G-LR-S Upon reload or power-cycle, will see interfaces on the switch go down or errdisable due to sfp-config-mismatch If interface is down or errdisabled can check if there is mismatch in hardware speed: Switch#show platform hardware fed switch active npu slot 1 port 12 port-info ----------- Display L1 port debug info------------ |-------------------------------------------------| | SW port debug states | |-------------------------------------------------| | Slot/Port | 1/12 | Speed | speed_25Gbps <<<<<< Checking what this is set to. If interface is configured for 10G, but 25G is seen here; that is indicative of this defect. | Active | Yes | XCVR Presence | Present | Admin | Enabled | Disabled | Yes | Link State | Down | Init Done | Yes | XCVR Type | 750 | Breakout | Disabled | Delete Mask | 0 | LED State | Amber On | Breakout Progress | 0 | Breakout Mode | 7 | Maximum Subports | 0 | Port Group Cfg | 2 | Port Group mode | 0 | Flow Control Mode | TX/RX OFF |-------------------------------------------------| |-------------------------------------------------| | LED port debug states | |-------------------------------------------------| | Slot/Port | 1/12 | Suspended | Yes | Stp blocked | No | Txrx_error | No | Txrx_active | No |-------------------------------------------------| LC present 0 Outstanding ports to be deleted 48 Total failed deleted ports 0 Check what ports are in port-group: Switch#show hw-module switch 1 slot 1 port-group speed ------------------ show portmap switch 1 slot 1 speed ------------------ -------------------------------------------------- Port-groupPorts Speed -------------------------------------------------- 1 Fif1/0/1, Fif1/0/2 NONE 2 Fif1/0/3, Fif1/0/4 NONE 3 Fif1/0/5, Fif1/0/6 NONE 4 Fif1/0/7, Fif1/0/8 NONE 5 Fif1/0/9, Fif1/0/10 NONE 6 Fif1/0/11, Fif1/0/12 NONE

Workaround

1. Configure 'Errdisable detect cause sfp-config-mismatch' and 'Errdisable Recovery cause sfp-config-mismatch'. This allows the errdisable state to be detected and the switch will attempt to recover the interface after set amount of time. The default time for errdisable recovery is 5 minutes (300 seconds); this can be changed down to 30 seconds with command 'errdisable recovery interval 30'. Switch(config)#errdisable recovery interval 30 2. Shut / No shut the affected interfaces if they are errdisable and errdisable recovery is not configured. 3. Remove and re-insert the optics of the affected interfaces. 4. Use hard-coded optics instead of dual-rate. For example, SFP-10G-LR is a hard-coded optic, and SFP-10/25G-LR is a dual-rate optic.

Further Problem Description