Linux ethtool Statistics - UG1739

Number of packets dropped by the network adapter because of a lack of RX descriptors in the RX queue.

Packets can be dropped by the NIC when there are insufficient RX descriptors in the RX queue to allocate to the packet. This problem occurs if the receive rate is very high and the network adapter receive cycle process has insufficient time between processing to refill the queue with new descriptors.

A number of different steps can be tried to resolve this issue:

• Disable the irqbalance daemon in the OS
• Distribute the traffic load across the available CPU/cores by setting rss_cpus=cores. Refer to Receive Side Scaling section
• Increase receive queue size using ethtool.

• Packet rate exceeds maximum supported by the adapter.
• Adapter is inserted into a low speed or low width PCI slot – so the PCIe bus cannot support the required bandwidth.
• Packets are being replicated by the adapter and the resulting bandwidth cannot be handled by the PCIe bus.
• Host memory bandwidth is being used by other devices resulting in poor performance for the adapter.

• The packet does not match any filter.
• The matched filter indicates the packet must be dropped.

When the host fails to push packet bytes fast enough to match the adapter port speed.

The packet is truncated and data transmitted as a poisoned packet.

When host fails to send all bytes to complete the packet to be sent by CTPIO before the VI inactivity timer expires.

The packet is truncated and data transmitted as a poisoned packet.

A non-sequential address (for packet data) is encountered during CTPIO, caused when packet data is sent over PCIe interface as out-of-order or with gaps.

Packet is truncated and transmitted as a poisoned packet.

If packet length is less than length advertised in the CTPIO header the CTPIO fails.

Or packet write is not aligned to (or multiple of) 32-bytes.

The packet might be transmitted as a poisoned packet if sending has already started, or be erased if send has not already started.

When the VI is not enabled to send using CTPIO, or first write is not the packet header.

The packet is sent using DMA datapath.

Length in header < 29 bytes.

The packet is sent using DMA datapath.

Number of instances when CTPIO push was rejected. This can occur because:

• The VI legacy datapath is still busy.
• Another CTPIO is in progress.
• VI is not enabled to use CTPIO.
• Push request for illegal sized frame.

Fallback events do not result in poison packets. Rejected packets use the DMA datapath path.

When the packet send has started, if CTPIO has to abort this packet, a corrupt CRC is attached to the packet.

A poisoned packet can be sent over the wire, depending on the mode.

The packet is sent using DMA datapath.

Before a packet send has started. Corrupt, undersized or poisoned packets are erased from the CTPIO datapath.

The packet is sent using DMA datapath.

• Mismatch unicast address and unicast promiscuous mode is not enabled.
• Packet is a pause frame.
• Packet has length discrepancy.
• Due to internal FIFO overflow condition.
• Length < 60 bytes.

Number of packets dropped by the adapter when there are insufficient socket buffers available to receive packets into.

See also port_rx_nodesc_drop_cnt and port_rx_nodesc_drops above.

These PTP stats counters relate to the mechanism used by sfptpd to synchronize the system clock and adapter clock(s) in a server.

For each synchronization event, sfptpd selects a number of system clock times to be compared to the adapter clock time. If the times can be synchronized, the good_syncs counter is incremented, otherwise the bad_syncs counter is incremented. If sfptpd is unable to synchronize the clocks at this event, the sync_timeout counter is incremented.

sfptpd synchronizes clocks 16 times per second, so incrementing counters do not necessarily indicate bad synchronization between local server clocks and an external PTP master clock.