Cannot get "nvlink_flit_crc_error_count_total(409)" and "nvlink_data_crc_error_count_total(419)" in H800 System

[zdyang]

No description provided.

[zdyang]

I used dcgm 3.3.0 in a H100 system with nvlink and tried to collect nvlink related metrics, such as "nvlink_flit_crc_error_count_total"(field id 409) and "nvlink_data_crc_error_count_total" (field id 419). However, dcgm always returned N/A. nv-hostengine and dcgmi was run with root privilege. Hareware driver version was "535.129"

Querying with dcgmi always returned "na"

[zdyang]

And the log from. "/var/nv-hostengine".

DEBUG [2064103:2066860] Appended entity i64 eg 1, eid 7, fieldId 409, ts 1716390313213008, value1 9223372036854775794, value2 0, cached 1, buffered 0 [/workspaces/dcgm-rel_dcgm_3_3-postmerge/dcgmlib/src/DcgmCacheManager.cpp:6260] [DcgmCacheManager::AppendEntityInt64]