Data Domain: Intel Interfaces link down with tx_timeout

Zhrnutie: Data domains with intel cards interface links may go down with tx_timeout recovery unsuccessful, device is in an unrecoverable state.

Tento článok sa vzťahuje na Tento článok sa nevzťahuje na Tento článok nie je viazaný na žiadny konkrétny produkt. V tomto článku nie sú uvedené všetky verzie produktov.

Symptómy

The following error logs are found in kern.info. 

kernel: [13023278.800638][T14886] (E4)irdma: probe of ice.roce.5 failed with error -110 
kernel: [13023278.800833][    C9] (E4)ice 0000:ae:00.0 eth8a: tx_timeout: VSI_num: 8, Q 41, NTC: 0x1bd, HW_HEAD: 0x1cb, NTU: 0x1cc, INT: 0x4000000 
kernel: [13023278.800834][    C9] (E4)ice 0000:ae:00.0 eth8a: tx_timeout recovery level 1, txqueue 41 
kernel: [13022896.344077][    C7] (E4)ice 0000:0b:00.0 eth2a: tx_timeout recovery unsuccessful, device is in unrecoverable state.

You may view the kern.info logs with the following command: 

log view debug/platform/kern.info

You can also check in the support bundle by navigating to platform and grepping the error logs above. 

Príčina

There is an issue in the Intel irdma driver within the irdma_wait_pe_ready function. When RDMA is disabled in the BIOS, the function performs a spin-wait for up to 15 seconds. This prolonged spin-wait can trigger RCU stalls, preventing some CPUs from being scheduled to handle TX and RX interrupts. As a result, NIC transmit and receive operations may time out. This issue occurs during the system reboot phase.

Riešenie

Upgrade to DDOS 8.4 or later versions. 

The fix reduces the spin-wait duration from 15 to 1.5 seconds, eliminating the problem.

Dotknuté produkty

Data Domain
Vlastnosti článku
Číslo článku: 000478096
Typ článku: Solution
Dátum poslednej úpravy: 16 jún 2026
Verzia:  2
Nájdite odpovede na svoje otázky od ostatných používateľov spoločnosti Dell
Služby podpory
Skontrolujte, či sa na vaše zariadenie vzťahujú služby podpory.