Technical Tip: Troubleshooting NP6XLite PDQ queue stuck causing packet drops

Description

This article describes how to verify whether queues inside NP6XLite are stuck or building backlog, which may result in packet drops or latency.

Scope

FortiGate models with NP6XLite.

Solution

Run the following command:

fnsysctl cat /proc/net/np6xlite_0/pdq IHP->ISW    00003dd6    00003dd6 HTX->ISW    000046ad    000046ad LPBK->ISW   00000000    00000000 ISW->SSE    0000b2a6    0000b2a6 SSE_LBLC    0000ccc6    0000ccc6 SSE->OSW    0000b2a6    0000b2a6 PNM->OSW    00000000    00000000 OS->EHP    000046ad    000046ad HRX_TUNPDQ  000015d4    000015d4 IPSEC_I     00000481    00000481 IPSEC_O     000029a2    000029a2 IPT_I       00000000    00000000 IPT_O       00000000    00000000 IPTO->PIPE  00000000    00000000 CWI         00000000    00000000 CWO         00000000    00000000 CWO>PIPE   00000000    00000000 OSW->PPF    00000000    00000000 PPF_PIPE    00000000    00000000 SYNK        00000000    00000000

Run this command multiple times at short intervals to track whether counters are increasing or queues are stalled.

Output meaning.

Each line represents a PDQ queue between two processing stages inside NP6XLite.

Fields:

• wp_cnt (Write Pointer Count) – packets written into the queue.
• rp_cnt (Read Pointer Count) – packets read from the queue.

Interpretation:

• If wp_cnt == rp_cnt → Queue is healthy, packets are flowing.
• If wp_cnt > rp_cnt → Queue backlog, potential packet drops due to overflow.
• If wp_cnt < rp_cnt → Abnormal, possible counter reset or error (should be investigated).