perf(memory): add reclaim signals and cache controls (#2689)

This commit is contained in:
houseme
2026-04-27 00:42:35 +08:00
committed by GitHub
parent 438913b84a
commit 50d03ef021
23 changed files with 2253 additions and 109 deletions
@@ -2534,7 +2534,7 @@
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_system_process_resident_memory_bytes{job=~\"$job\"})",
"expr": "sum by (job) (rustfs_system_process_resident_memory_bytes{job=~\"$job\"}) or sum by (job) (rustfs_memory_process_resident_bytes{job=~\"$job\"})",
"legendFormat": "process rss - {{job}}",
"range": true,
"refId": "A"
@@ -2545,8 +2545,8 @@
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "rustfs_system_memory_used_bytes{job=~\"$job\"}",
"legendFormat": "system used - {{job}}",
"expr": "sum by (job) (rustfs_memory_process_virtual_bytes{job=~\"$job\"})",
"legendFormat": "process virtual - {{job}}",
"range": true,
"refId": "B"
},
@@ -2556,13 +2556,755 @@
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "rustfs_system_memory_used_perc{job=~\"$job\"}",
"legendFormat": "system used percent - {{job}}",
"expr": "sum by (job) (rustfs_memory_cgroup_anon_bytes{job=~\"$job\"})",
"legendFormat": "anon split - {{job}}",
"range": true,
"refId": "C"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_cgroup_file_bytes{job=~\"$job\"})",
"legendFormat": "file split - {{job}}",
"range": true,
"refId": "D"
}
],
"title": "Memory Split",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 10,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "never",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 12,
"x": 0,
"y": 80
},
"id": 501,
"options": {
"legend": {
"calcs": [
"lastNotNull",
"max"
],
"displayMode": "table",
"placement": "right",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "12.3.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_delete_tail_activity_total_inflight_current{job=~\"$job\"})",
"legendFormat": "delete tail inflight - {{job}}",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_allocator_reclaim_scanner_activity_current{job=~\"$job\"})",
"legendFormat": "scanner activity - {{job}}",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_allocator_reclaim_heal_activity_current{job=~\"$job\"})",
"legendFormat": "heal activity - {{job}}",
"range": true,
"refId": "C"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_allocator_reclaim_reclaimable_work_current{job=~\"$job\"})",
"legendFormat": "reclaimable work - {{job}}",
"range": true,
"refId": "D"
}
],
"title": "Tail / Reclaim Activity",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 10,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "never",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 12,
"x": 12,
"y": 80
},
"id": 502,
"options": {
"legend": {
"calcs": [
"lastNotNull",
"max"
],
"displayMode": "table",
"placement": "right",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "12.3.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_allocator_reclaim_idle_streak{job=~\"$job\"})",
"legendFormat": "idle streak - {{job}}",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rate(rustfs_memory_allocator_reclaim_total{job=~\"$job\",result=\"ok\"}[5m]))",
"legendFormat": "reclaim ok rate - {{job}}",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job,reason) (rate(rustfs_memory_allocator_reclaim_skipped_total{job=~\"$job\"}[5m]))",
"legendFormat": "reclaim skipped {{reason}} - {{job}}",
"range": true,
"refId": "C"
}
],
"title": "Memory Usage",
"title": "Allocator Reclaim Health",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 10,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "never",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "bytes"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 12,
"x": 0,
"y": 87
},
"id": 503,
"options": {
"legend": {
"calcs": [
"lastNotNull",
"max"
],
"displayMode": "table",
"placement": "right",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "12.3.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_cgroup_current_bytes{job=~\"$job\"})",
"legendFormat": "cgroup current - {{job}}",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_cgroup_limit_bytes{job=~\"$job\"})",
"legendFormat": "cgroup limit - {{job}}",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_cgroup_active_file_bytes{job=~\"$job\"})",
"legendFormat": "active file - {{job}}",
"range": true,
"refId": "C"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_cgroup_inactive_file_bytes{job=~\"$job\"})",
"legendFormat": "inactive file - {{job}}",
"range": true,
"refId": "D"
}
],
"title": "Memory Cgroup Detail",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 10,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "never",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 12,
"x": 12,
"y": 87
},
"id": 504,
"options": {
"legend": {
"calcs": [
"lastNotNull",
"max"
],
"displayMode": "table",
"placement": "right",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "12.3.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_ec_encode_inflight_bytes_current{job=~\"$job\"})",
"legendFormat": "ec inflight bytes - {{job}}",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_get_object_buffered_bytes_current{job=~\"$job\"})",
"legendFormat": "get buffered bytes - {{job}}",
"range": true,
"refId": "B"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_allocator_reclaim_active_requests{job=~\"$job\"})",
"legendFormat": "allocator active requests - {{job}}",
"range": true,
"refId": "C"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job) (rustfs_memory_allocator_reclaim_delete_tail_activity_current{job=~\"$job\"})",
"legendFormat": "allocator tail activity - {{job}}",
"range": true,
"refId": "D"
}
],
"title": "Heap Amplification Signals",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 10,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "never",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "short"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 12,
"x": 0,
"y": 94
},
"id": 505,
"options": {
"legend": {
"calcs": [
"lastNotNull",
"max"
],
"displayMode": "table",
"placement": "right",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "12.3.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job,stage) (rustfs_delete_tail_activity_inflight_current{job=~\"$job\"})",
"legendFormat": "{{stage}} inflight - {{job}}",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job,stage) (rate(rustfs_delete_tail_activity_started_total{job=~\"$job\"}[5m]))",
"legendFormat": "{{stage}} started rate - {{job}}",
"range": true,
"refId": "B"
}
],
"title": "Delete Tail by Stage",
"type": "timeseries"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"fieldConfig": {
"defaults": {
"color": {
"mode": "palette-classic"
},
"custom": {
"axisBorderShow": false,
"axisCenteredZero": false,
"axisColorMode": "text",
"axisLabel": "",
"axisPlacement": "auto",
"barAlignment": 0,
"barWidthFactor": 0.6,
"drawStyle": "line",
"fillOpacity": 10,
"gradientMode": "none",
"hideFrom": {
"legend": false,
"tooltip": false,
"viz": false
},
"insertNulls": false,
"lineInterpolation": "smooth",
"lineWidth": 1,
"pointSize": 5,
"scaleDistribution": {
"type": "linear"
},
"showPoints": "never",
"showValues": false,
"spanNulls": false,
"stacking": {
"group": "A",
"mode": "none"
},
"thresholdsStyle": {
"mode": "off"
}
},
"mappings": [],
"thresholds": {
"mode": "absolute",
"steps": [
{
"color": "green",
"value": 0
}
]
},
"unit": "bytes"
},
"overrides": []
},
"gridPos": {
"h": 7,
"w": 12,
"x": 12,
"y": 94
},
"id": 506,
"options": {
"legend": {
"calcs": [
"lastNotNull",
"max"
],
"displayMode": "table",
"placement": "right",
"showLegend": true
},
"tooltip": {
"hideZeros": false,
"mode": "multi",
"sort": "desc"
}
},
"pluginVersion": "12.3.2",
"targets": [
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job,kind,result) (rate(rustfs_page_cache_reclaim_requests_total{job=~\"$job\"}[5m]))",
"legendFormat": "{{kind}} reclaim {{result}} rate - {{job}}",
"range": true,
"refId": "A"
},
{
"datasource": {
"type": "prometheus",
"uid": "${datasource}"
},
"editorMode": "code",
"expr": "sum by (job,kind) (rate(rustfs_page_cache_reclaim_bytes_total{job=~\"$job\"}[5m]))",
"legendFormat": "{{kind}} reclaim bytes/s - {{job}}",
"range": true,
"refId": "B"
}
],
"title": "Page Cache Reclaim",
"type": "timeseries"
},
{