代码改变世界

prometheus 告警 gpu 卡的ecc错误

2020-07-31 17:24  ZealouSnesS  阅读(912)  评论(0编辑  收藏  举报
prometheus有两个指标可以告警gpu卡的错误

# HELP dcgm_ecc_dbe_volatile_total Total number of double-bit volatile ECC errors. 双位易失性ECC错误的总数 # TYPE dcgm_ecc_dbe_volatile_total counter dcgm_ecc_dbe_volatile_total{gpu="0",uuid="GPU-4d52e430-b8c7-a0b9-7fda-4aa825af5c97"} 0 # HELP dcgm_ecc_sbe_volatile_total Total number of single-bit volatile ECC errors. 单位易失性ECC错误的总数 # TYPE dcgm_ecc_sbe_volatile_total counter dcgm_ecc_sbe_volatile_total{gpu="0",uuid="GPU-4d52e430-b8c7-a0b9-7fda-4aa825af5c97"} 0