Abnormal Disk I/O Persists on LACP Bond After Switch Port Reset and TCAM Misconfiguration
Applies to
- SolidFire Storage Nodes
- SolidFire H-Series Storage Nodes
- Element Software
- SFP: CISCO-FINISAR SFP-H10GB-AOC5M FCBG110SD1C05-CS
- Cisco Nexus 93180YC-FX
- Cisco ACI switch
- LACP rate Fast; LACP suspend individual off
- TCAM (Ternary Content Addressable Memory)
Issue
If TCAM is not properly configured on the Nexus 93180YC-FX switches, the SolidFire cluster may experience dropped packets across the virtual peer-link tunnel, potentially causing performance degradation or connectivity issues that impact application performance and storage access from the host perspective.
- Cisco Nexus 93180YC-FX vPC switches lack the required TCAM configuration
- LACP bond doesn't recover after bringing port down and then back up on a switch
solidfire_node ~ # cat /proc/net/bonding/Bond10G Active Aggregator Info: Aggregator ID: 1 Number of ports: 1 (Instead of 2)
- 10G link on SolidFire side remains UP when the port is Disabled (Administratively Down) on a switch side
