microsoft / pai / 16612 / 12

34%

master: 34%

push

travis-ci

add alert for gpu count changes (#3297)

Add alert when node GPU count is changed. This may caused by driver issue.

Add a record rule to generate a new metric: node:nvidiasmi_utilization_gpu:count, which is calculated by: count(nvidiasmi_utilization_gpu) by (instance)

Add a new alert when changes(node:nvidiasmi_utilization_gpu:count) > 0

517 of 1029 branches covered (50.24%)

Branch coverage included in aggregate %.

1691 of 2626 relevant lines covered (64.39%)

6.5 hits per line