The rx_crc_errors counter increments when frames arriving on the interface fail the Cyclic Redundancy Check, which means the frame's bits were altered in transit before reaching the NIC. This is a Layer 1 (physical) symptom, most commonly caused by a damaged or poor-quality cable, a failing or incompatible transceiver (e.g., SFP/twinax), a loose connector, or other physical-media issues along the link — including upstream cabling or switch-port problems, since modern data-center switches use cut-through forwarding and may pass along already-corrupted frames before completing their own CRC check. A duplex mismatch typically manifests as late collisions/runt frames rather than pure CRC errors, MTU issues would show up as length/fragmentation errors, and CPU saturation would show up as missed or FIFO errors rather than CRC corruption. This makes physical-layer corruption (faulty cable/transceiver) the most likely root cause of a steadily incrementing rx_crc_errors counter.
Community Discussion