This article describes steps to resolve TOE memory read failure on the QFX10002 platform.

Traffic blackhole might be observed, and ARP resolution might fail if the following TOE Error is seen

root@qfx10002> show chassis alarms no-forwarding
1 alarms currently active
Alarm time        Class Description
2023-05-24 12:46:01 UTC Major FPC 0 Major Errors - TOE Error code: 0x12c06

TOE memory errors will appear in the log messages:

[Wed May 24 13:56:51.447 LOG: Err] toe_ucode_read_obj_struct: failed to read obj for type 24 inst 0.
[Wed May 24 13:56:51.447 LOG: Err] hostif_pio_read_local_mem: PE-2 HOSTIF toe_addr 0x10000 over limit 0x10000
[Wed May 24 13:56:51.447 LOG: Err] Failed to read TOE memory at 0x3ffff for TOE 1: Size: 60 Index: 1.
[Wed May 24 13:56:51.447 LOG: Emergency] TOE memory read failure:

A device reboot is required to clear this error state. In certain cases, the FPC may not come online after the reboot and will need replacing.

Shutting down the affected PFE and/or interface associated with the affected PFE may decrease the impact on the transit traffic. However, locally generated traffic may still blackhole in the egress direction. For instance, ARP resolution might fail.