| 1 | # you can disable an alarm notification by setting the 'to' line to: silent |
| 2 | |
| 3 | # --- Quota --- |
| 4 | |
| 5 | template: am_ml_quota_utilization |
| 6 | on: azure_monitor.machine_learning.quota_utilization |
| 7 | class: Utilization |
| 8 | type: Other |
| 9 | component: Azure ML |
| 10 | lookup: average -5m unaligned of utilization |
| 11 | units: percentage |
| 12 | every: 1m |
| 13 | warn: $this > (($status >= $WARNING) ? (75) : (85)) |
| 14 | crit: $this > (($status == $CRITICAL) ? (85) : (95)) |
| 15 | delay: down 5m multiplier 1.5 max 1h |
| 16 | summary: ML quota utilization on ${label:resource_name} |
| 17 | info: Average compute quota utilization on Azure ML workspace ${label:resource_name} \ |
| 18 | in ${label:resource_group} (${label:region}). \ |
| 19 | High quota utilization may prevent new jobs from starting. |
| 20 | to: sysadmin |
| 21 | |
| 22 | # --- Cluster Cores --- |
| 23 | |
| 24 | template: am_ml_unusable_cores |
| 25 | on: azure_monitor.machine_learning.cluster_cores |
| 26 | class: Errors |
| 27 | type: Other |
| 28 | component: Azure ML |
| 29 | lookup: average -5m unaligned of unusable |
| 30 | units: cores |
| 31 | every: 1m |
| 32 | warn: $this > (($status >= $WARNING) ? (0) : (2)) |
| 33 | crit: $this > (($status == $CRITICAL) ? (2) : (10)) |
| 34 | delay: down 5m multiplier 1.5 max 1h |
| 35 | summary: ML unusable cores on ${label:resource_name} |
| 36 | info: Number of unusable compute cores on Azure ML workspace ${label:resource_name} \ |
| 37 | in ${label:resource_group} (${label:region}). \ |
| 38 | Unusable cores indicate hardware or configuration failures. |
| 39 | to: sysadmin |
| 40 | |
| 41 | template: am_ml_preempted_cores |
| 42 | on: azure_monitor.machine_learning.cluster_cores |
| 43 | class: Workload |
| 44 | type: Other |
| 45 | component: Azure ML |
| 46 | lookup: average -5m unaligned of preempted |
| 47 | units: cores |
| 48 | every: 1m |
| 49 | warn: $this > (($status >= $WARNING) ? (5) : (10)) |
| 50 | delay: down 5m multiplier 1.5 max 1h |
| 51 | summary: ML preempted cores on ${label:resource_name} |
| 52 | info: Number of preempted compute cores on Azure ML workspace ${label:resource_name} \ |
| 53 | in ${label:resource_group} (${label:region}). \ |
| 54 | Preempted cores may cause training job interruptions. |
| 55 | to: sysadmin |
| 56 | |
| 57 | # --- Cluster Nodes --- |
| 58 | |
| 59 | template: am_ml_unusable_nodes |
| 60 | on: azure_monitor.machine_learning.cluster_nodes |
| 61 | class: Errors |
| 62 | type: Other |
| 63 | component: Azure ML |
| 64 | lookup: average -5m unaligned of unusable |
| 65 | units: nodes |
| 66 | every: 1m |
| 67 | warn: $this > (($status >= $WARNING) ? (0) : (1)) |
| 68 | crit: $this > (($status == $CRITICAL) ? (1) : (5)) |
| 69 | delay: down 5m multiplier 1.5 max 1h |
| 70 | summary: ML unusable nodes on ${label:resource_name} |
| 71 | info: Number of unusable compute nodes on Azure ML workspace ${label:resource_name} \ |
| 72 | in ${label:resource_group} (${label:region}). \ |
| 73 | Unusable nodes indicate hardware or configuration failures. |
| 74 | to: sysadmin |
| 75 | |
| 76 | # --- CPU Utilization --- |
| 77 | |
| 78 | template: am_ml_cpu_utilization |
| 79 | on: azure_monitor.machine_learning.cpu_utilization |
| 80 | class: Utilization |
| 81 | type: Other |
| 82 | component: Azure ML |
| 83 | lookup: average -5m unaligned of cluster_cpu |
| 84 | units: percentage |
| 85 | every: 1m |
| 86 | warn: $this > (($status >= $WARNING) ? (80) : (90)) |
| 87 | crit: $this > (($status == $CRITICAL) ? (90) : (95)) |
| 88 | delay: down 5m multiplier 1.5 max 1h |
| 89 | summary: ML CPU utilization on ${label:resource_name} |
| 90 | info: Average cluster CPU utilization on Azure ML workspace ${label:resource_name} \ |
| 91 | in ${label:resource_group} (${label:region}) |
| 92 | to: sysadmin |
| 93 | |
| 94 | # --- CPU Memory Utilization --- |
| 95 | |
| 96 | template: am_ml_cpu_memory_utilization |
| 97 | on: azure_monitor.machine_learning.cpu_memory_utilization |
| 98 | class: Utilization |
| 99 | type: Other |
| 100 | component: Azure ML |
| 101 | lookup: average -5m unaligned of utilization |
| 102 | units: percentage |
| 103 | every: 1m |
| 104 | warn: $this > (($status >= $WARNING) ? (80) : (90)) |
| 105 | crit: $this > (($status == $CRITICAL) ? (90) : (95)) |
| 106 | delay: down 5m multiplier 1.5 max 1h |
| 107 | summary: ML CPU memory utilization on ${label:resource_name} |
| 108 | info: Average CPU memory utilization on Azure ML workspace ${label:resource_name} \ |
| 109 | in ${label:resource_group} (${label:region}) |
| 110 | to: sysadmin |
| 111 | |
| 112 | # --- GPU Utilization --- |
| 113 | |
| 114 | template: am_ml_gpu_utilization |
| 115 | on: azure_monitor.machine_learning.gpu_utilization |
| 116 | class: Utilization |
| 117 | type: Other |
| 118 | component: Azure ML |
| 119 | lookup: average -5m unaligned of cluster_gpu |
| 120 | units: percentage |
| 121 | every: 1m |
| 122 | warn: $this > (($status >= $WARNING) ? (85) : (95)) |
| 123 | crit: $this > (($status == $CRITICAL) ? (95) : (99)) |
| 124 | delay: down 5m multiplier 1.5 max 1h |
| 125 | summary: ML GPU utilization on ${label:resource_name} |
| 126 | info: Average cluster GPU utilization on Azure ML workspace ${label:resource_name} \ |
| 127 | in ${label:resource_group} (${label:region}) |
| 128 | to: sysadmin |
| 129 | |
| 130 | # --- GPU Memory Utilization --- |
| 131 | |
| 132 | template: am_ml_gpu_memory_utilization |
| 133 | on: azure_monitor.machine_learning.gpu_memory_utilization |
| 134 | class: Utilization |
| 135 | type: Other |
| 136 | component: Azure ML |
| 137 | lookup: average -5m unaligned of cluster_gpu_memory |
| 138 | units: percentage |
| 139 | every: 1m |
| 140 | warn: $this > (($status >= $WARNING) ? (80) : (90)) |
| 141 | crit: $this > (($status == $CRITICAL) ? (90) : (95)) |
| 142 | delay: down 5m multiplier 1.5 max 1h |
| 143 | summary: ML GPU memory utilization on ${label:resource_name} |
| 144 | info: Average cluster GPU memory utilization on Azure ML workspace ${label:resource_name} \ |
| 145 | in ${label:resource_group} (${label:region}). \ |
| 146 | High GPU memory usage may cause out-of-memory training failures. |
| 147 | to: sysadmin |
| 148 | |
| 149 | # --- Disk Usage --- |
| 150 | |
| 151 | template: am_ml_disk_utilization |
| 152 | on: azure_monitor.machine_learning.disk_usage |
| 153 | class: Utilization |
| 154 | type: Other |
| 155 | component: Azure ML |
| 156 | calc: ($used + $available > 0) ? ($used * 100 / ($used + $available)) : (0) |
| 157 | units: % |
| 158 | every: 1m |
| 159 | warn: $this > (($status >= $WARNING) ? (75) : (85)) |
| 160 | crit: $this > (($status == $CRITICAL) ? (85) : (95)) |
| 161 | delay: down 5m multiplier 1.5 max 1h |
| 162 | summary: ML disk utilization on ${label:resource_name} |
| 163 | info: Disk utilization percentage on Azure ML workspace ${label:resource_name} \ |
| 164 | in ${label:resource_group} (${label:region}). \ |
| 165 | High disk usage can cause training job failures. |
| 166 | to: sysadmin |
| 167 | |
| 168 | # --- Model Deployments --- |
| 169 | |
| 170 | template: am_ml_model_deploy_failures |
| 171 | on: azure_monitor.machine_learning.model_deployments |
| 172 | class: Errors |
| 173 | type: Other |
| 174 | component: Azure ML |
| 175 | lookup: sum -5m unaligned of failed |
| 176 | units: deployments |
| 177 | every: 1m |
| 178 | warn: $this > (($status >= $WARNING) ? (0) : (1)) |
| 179 | crit: $this > (($status == $CRITICAL) ? (1) : (5)) |
| 180 | delay: down 5m multiplier 1.5 max 1h |
| 181 | summary: ML model deployment failures on ${label:resource_name} |
| 182 | info: Number of failed model deployments over the last 5 minutes on Azure ML workspace ${label:resource_name} \ |
| 183 | in ${label:resource_group} (${label:region}) |
| 184 | to: sysadmin |
| 185 | |
| 186 | # --- Model Registrations --- |
| 187 | |
| 188 | template: am_ml_model_register_failures |
| 189 | on: azure_monitor.machine_learning.model_registrations |
| 190 | class: Errors |
| 191 | type: Other |
| 192 | component: Azure ML |
| 193 | lookup: sum -5m unaligned of failed |
| 194 | units: registrations |
| 195 | every: 1m |
| 196 | warn: $this > (($status >= $WARNING) ? (0) : (1)) |
| 197 | crit: $this > (($status == $CRITICAL) ? (1) : (5)) |
| 198 | delay: down 5m multiplier 1.5 max 1h |
| 199 | summary: ML model registration failures on ${label:resource_name} |
| 200 | info: Number of failed model registrations over the last 5 minutes on Azure ML workspace ${label:resource_name} \ |
| 201 | in ${label:resource_group} (${label:region}) |
| 202 | to: sysadmin |
| 203 | |
| 204 | # --- Run Completion --- |
| 205 | |
| 206 | template: am_ml_failed_runs |
| 207 | on: azure_monitor.machine_learning.run_completion |
| 208 | class: Errors |
| 209 | type: Other |
| 210 | component: Azure ML |
| 211 | lookup: sum -5m unaligned of failed |
| 212 | units: runs |
| 213 | every: 1m |
| 214 | warn: $this > (($status >= $WARNING) ? (0) : (3)) |
| 215 | crit: $this > (($status == $CRITICAL) ? (3) : (10)) |
| 216 | delay: down 5m multiplier 1.5 max 1h |
| 217 | summary: ML failed runs on ${label:resource_name} |
| 218 | info: Number of failed training/experiment runs over the last 5 minutes on Azure ML workspace ${label:resource_name} \ |
| 219 | in ${label:resource_group} (${label:region}) |
| 220 | to: sysadmin |
| 221 | |
| 222 | template: am_ml_not_responding_runs |
| 223 | on: azure_monitor.machine_learning.run_completion |
| 224 | class: Availability |
| 225 | type: Other |
| 226 | component: Azure ML |
| 227 | lookup: sum -5m unaligned of not_responding |
| 228 | units: runs |
| 229 | every: 1m |
| 230 | warn: $this > (($status >= $WARNING) ? (0) : (1)) |
| 231 | crit: $this > (($status == $CRITICAL) ? (1) : (5)) |
| 232 | delay: down 5m multiplier 1.5 max 1h |
| 233 | summary: ML not-responding runs on ${label:resource_name} |
| 234 | info: Number of runs that stopped responding over the last 5 minutes on Azure ML workspace ${label:resource_name} \ |
| 235 | in ${label:resource_group} (${label:region}). \ |
| 236 | Not-responding runs indicate compute or infrastructure issues. |
| 237 | to: sysadmin |
| 238 | |
| 239 | # --- Run Issues --- |
| 240 | |
| 241 | template: am_ml_run_errors |
| 242 | on: azure_monitor.machine_learning.run_issues |
| 243 | class: Errors |
| 244 | type: Other |
| 245 | component: Azure ML |
| 246 | lookup: sum -5m unaligned of errors |
| 247 | units: errors |
| 248 | every: 1m |
| 249 | warn: $this > (($status >= $WARNING) ? (0) : (5)) |
| 250 | crit: $this > (($status == $CRITICAL) ? (5) : (20)) |
| 251 | delay: down 5m multiplier 1.5 max 1h |
| 252 | summary: ML run errors on ${label:resource_name} |
| 253 | info: Number of run errors over the last 5 minutes on Azure ML workspace ${label:resource_name} \ |
| 254 | in ${label:resource_group} (${label:region}) |
| 255 | to: sysadmin |
| 256 | |
| 257 | # --- Storage API --- |
| 258 | |
| 259 | template: am_ml_storage_api_total |
| 260 | on: azure_monitor.machine_learning.storage_api_calls |
| 261 | class: Workload |
| 262 | type: Other |
| 263 | component: Azure ML |
| 264 | lookup: sum -5m unaligned of success,failure |
| 265 | units: calls |
| 266 | every: 1m |
| 267 | info: Total storage API calls over the last 5 minutes on Azure ML workspace ${label:resource_name} |
| 268 | |
| 269 | template: am_ml_storage_api_failures |
| 270 | on: azure_monitor.machine_learning.storage_api_calls |
| 271 | class: Errors |
| 272 | type: Other |
| 273 | component: Azure ML |
| 274 | lookup: sum -5m unaligned of failure |
| 275 | calc: ($am_ml_storage_api_total > 10) ? ($this * 100 / $am_ml_storage_api_total) : (0) |
| 276 | units: % |
| 277 | every: 1m |
| 278 | warn: $this > (($status >= $WARNING) ? (1) : (5)) |
| 279 | crit: $this > (($status == $CRITICAL) ? (5) : (15)) |
| 280 | delay: down 5m multiplier 1.5 max 1h |
| 281 | summary: ML storage API failure rate on ${label:resource_name} |
| 282 | info: Percentage of failed storage API calls over the last 5 minutes on Azure ML workspace ${label:resource_name} \ |
| 283 | in ${label:resource_group} (${label:region}) |
| 284 | to: sysadmin |