prometheus queries

# cpu usage per container
sum(rate(container_cpu_usage_seconds_total{pod_name=~"nginx.*"}[5m])* 100000) by (pod_name,namespace)/2500 > 10

# prometheus performance
rate(prometheus_tsdb_head_samples_appended_total[5m])
topk(10, count({job=~".+"}) by(__name__))

# Amount of projects we can deploy using 22GB memory request
sum(node_memory_MemFree_bytes{node=~"compute-.*"} / 1024 / 1024 / 1024) / 22

# Available memory on compute nodes
sum(node_memory_MemFree_bytes{node=~"compute-.*"} / 1024 / 1024 / 1024)

# Memory usage/limit/request per project
sum(container_memory_usage_bytes{namespace="test1"}) / 1024 / 1024 / 1024
sum(kube_pod_container_resource_limits_memory_bytes{namespace="test1"}) / 1024 / 1024 / 1024
sum(kube_pod_container_resource_requests_memory_bytes{namespace="test1"}) / 1024 / 1024 / 1024

# Memory limit/request percent
sum(kube_pod_container_resource_limits_memory_bytes{node=~'compute.*'}) / sum(node_memory_MemTotal{node=~'compute.*'}) * 100
sum(kube_pod_container_resource_requests_memory_bytes{node=~'compute.*'}) / sum(node_memory_MemTotal{node=~'compute.*'}) * 100

# order by memory usage
sort_desc(container_memory_usage_bytes{instance=~"compute-.*"} / 1024 / 1024 / 1024 ) > 1

# average memory usage 1 week
avg(avg_over_time(container_memory_usage_bytes{pod_name=~"nginx.*",container_name="nginx"}[1w]) / 1024 / 1024)

# check if fluentd is sending logs
sum(irate(container_network_transmit_bytes_total{pod_name=~"logging-fluentd.*"}[5m]) == 0) by (pod_name)