bstein/titan-iac
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

monitoring: fix hottest stats and gpu share

Browse Source
This commit is contained in:
Brad Stein 2025-11-17 23:40:22 -03:00
parent f4dd1de43f
commit aef3176c1c
3 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
scripts/render_dashboards.py
View File

@ -145,7 +145,7 @@ def astreae_free_expr(mount):
def topk_with_node(expr): def topk_with_node(expr):
return f'label_replace(topk(1, {expr}), "__name__", "$1", "node", "(.*)")' return f"topk(1, {expr})"
def node_net_expr(scope=""): def node_net_expr(scope=""):

8
services/monitoring/dashboards/atlas-overview.json
View File

@ -438,7 +438,7 @@
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) (((1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m]))) * 100) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) (((1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m]))) * 100) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true
@ -511,7 +511,7 @@
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) ((avg by (instance) ((node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) / node_memory_MemTotal_bytes * 100)) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) ((avg by (instance) ((node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) / node_memory_MemTotal_bytes * 100)) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true
@ -584,7 +584,7 @@
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) ((sum by (instance) (rate(node_network_receive_bytes_total{device!~\"lo\"}[5m]) + rate(node_network_transmit_bytes_total{device!~\"lo\"}[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) ((sum by (instance) (rate(node_network_receive_bytes_total{device!~\"lo\"}[5m]) + rate(node_network_transmit_bytes_total{device!~\"lo\"}[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true
@ -653,7 +653,7 @@
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) ((sum by (instance) (rate(node_disk_read_bytes_total[5m]) + rate(node_disk_written_bytes_total[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) ((sum by (instance) (rate(node_disk_read_bytes_total[5m]) + rate(node_disk_written_bytes_total[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true

8
services/monitoring/grafana-dashboard-overview.yaml
View File

@ -447,7 +447,7 @@ data:
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) (((1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m]))) * 100) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) (((1 - avg by (instance) (rate(node_cpu_seconds_total{mode=\"idle\"}[5m]))) * 100) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true
@ -520,7 +520,7 @@ data:
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) ((avg by (instance) ((node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) / node_memory_MemTotal_bytes * 100)) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) ((avg by (instance) ((node_memory_MemTotal_bytes - node_memory_MemAvailable_bytes) / node_memory_MemTotal_bytes * 100)) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true
@ -593,7 +593,7 @@ data:
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) ((sum by (instance) (rate(node_network_receive_bytes_total{device!~\"lo\"}[5m]) + rate(node_network_transmit_bytes_total{device!~\"lo\"}[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) ((sum by (instance) (rate(node_network_receive_bytes_total{device!~\"lo\"}[5m]) + rate(node_network_transmit_bytes_total{device!~\"lo\"}[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true
@ -662,7 +662,7 @@ data:
}, },
"targets": [ "targets": [
{ {
"expr": "label_replace(topk(1, avg by (node) ((sum by (instance) (rate(node_disk_read_bytes_total[5m]) + rate(node_disk_written_bytes_total[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\"))), \"__name__\", \"$1\", \"node\", \"(.*)\")", "expr": "topk(1, avg by (node) ((sum by (instance) (rate(node_disk_read_bytes_total[5m]) + rate(node_disk_written_bytes_total[5m]))) * on(instance) group_left(node) label_replace(node_uname_info{nodename!=\"\"}, \"node\", \"$1\", \"nodename\", \"(.*)\")))",
"refId": "A", "refId": "A",
"legendFormat": "{{node}}", "legendFormat": "{{node}}",
"instant": true "instant": true