Interface Health
by @vahagn-madatyan
Interface and link health assessment with error counter analysis, optical power monitoring, discard diagnosis, and utilization trending. Multi-vendor coverag...
CRC Errors on Fiber with Normal Optical Power
Optical power within spec but CRC errors incrementing. Common causes: wavelength mismatch between SFP types (e.g., SX connected to LR), dirty connector on the inside of the SFP cage (not the fiber tip), or SFP incompatibility with the switch (non-qualified optic). Try: clean the SFP receptacle, verify both ends use the same SFP type, test with a vendor-qualified optic.
Output Discards with Low Utilization
Interface shows <30% average utilization but output discards are climbing. This is almost always microburst traffic β sub-second spikes that exceed link capacity during the burst but average out below the utilization threshold. Diagnose with: per-queue drop counters (shows which traffic class), interface buffer allocation stats. Remediate with: QoS scheduling adjustments, increased interface buffer depth, or traffic shaping at the ingress point.
Interface Stuck in Down/Down After Cable Swap
Admin up, operationally down after replacing a cable or SFP. Check: SFP is
fully seated (push firmly until click), fiber polarity is correct (Tx-to-Rx
crossover), SFP type matches remote end, speed/duplex is set to auto or
matches. On [Cisco], check show interfaces [intf] | include err-disabled
β the port may have been error-disabled by a protection feature (BPDU guard,
UDLD, link-flap detection). Recover with shutdown / no shutdown after
fixing the root cause.
Flapping Interface with No Errors
Interface cycles up/down every few seconds with zero error counters. This suggests a negotiation or protocol issue, not a physical fault. Common causes: auto-negotiation incompatibility (force speed/duplex on both ends), STP topology changes causing repeated blocking/forwarding transitions, UDLD aggressive mode detecting unidirectional link. Check spanning-tree state and UDLD status on the interface.
Rising Laser Bias with Stable Tx Power
Laser bias current increasing over weeks/months while Tx power remains stable. The SFP is compensating for laser degradation by driving more current. This is normal aging but indicates the SFP will eventually fail β Tx power will drop when the laser can no longer compensate. Plan proactive replacement before the Tx power begins declining. Track the trend: if bias current exceeds 80% of the manufacturer's max specification, schedule replacement within 30 days.
clawhub install interface-health