| 1 | # you can disable an alarm notification by setting the 'to' line to: silent |
| 2 | |
| 3 | # --- Cognitive Services - Availability --- |
| 4 | |
| 5 | template: am_cognitive_services_availability |
| 6 | on: azure_monitor.cognitive_services.availability |
| 7 | class: Availability |
| 8 | type: Other |
| 9 | component: Cognitive Services |
| 10 | lookup: average -5m unaligned of availability |
| 11 | units: percentage |
| 12 | every: 1m |
| 13 | warn: $this < (($status >= $WARNING) ? (99.9) : (99)) |
| 14 | crit: $this < (($status == $CRITICAL) ? (99) : (90)) |
| 15 | delay: down 5m multiplier 1.5 max 1h |
| 16 | summary: Cognitive Services availability on ${label:resource_name} |
| 17 | info: Success rate of Cognitive Services resource ${label:resource_name} \ |
| 18 | in ${label:resource_group} (${label:region}) |
| 19 | to: sysadmin |
| 20 | |
| 21 | # --- Cognitive Services - Errors --- |
| 22 | |
| 23 | template: am_cognitive_services_server_errors |
| 24 | on: azure_monitor.cognitive_services.errors |
| 25 | class: Errors |
| 26 | type: Other |
| 27 | component: Cognitive Services |
| 28 | lookup: average -5m unaligned of server |
| 29 | units: errors/s |
| 30 | every: 1m |
| 31 | warn: $this > (($status >= $WARNING) ? (1) : (5)) |
| 32 | crit: $this > (($status == $CRITICAL) ? (5) : (20)) |
| 33 | delay: down 5m multiplier 1.5 max 1h |
| 34 | summary: Cognitive Services server errors on ${label:resource_name} |
| 35 | info: Rate of server errors (5xx) from Cognitive Services resource ${label:resource_name} \ |
| 36 | in ${label:resource_group} (${label:region}) |
| 37 | to: sysadmin |
| 38 | |
| 39 | template: am_cognitive_services_client_errors |
| 40 | on: azure_monitor.cognitive_services.errors |
| 41 | class: Errors |
| 42 | type: Other |
| 43 | component: Cognitive Services |
| 44 | lookup: average -5m unaligned of client |
| 45 | units: errors/s |
| 46 | every: 1m |
| 47 | warn: $this > (($status >= $WARNING) ? (10) : (50)) |
| 48 | delay: down 5m multiplier 1.5 max 1h |
| 49 | summary: Cognitive Services client errors on ${label:resource_name} |
| 50 | info: Rate of client errors (4xx) from Cognitive Services resource ${label:resource_name} \ |
| 51 | in ${label:resource_group} (${label:region}). \ |
| 52 | High client error rates may indicate misconfigured API calls or invalid requests |
| 53 | to: sysadmin |
| 54 | |
| 55 | # --- Cognitive Services - Latency --- |
| 56 | |
| 57 | template: am_cognitive_services_latency |
| 58 | on: azure_monitor.cognitive_services.latency |
| 59 | class: Latency |
| 60 | type: Other |
| 61 | component: Cognitive Services |
| 62 | lookup: average -5m unaligned of average |
| 63 | units: milliseconds |
| 64 | every: 1m |
| 65 | warn: $this > (($status >= $WARNING) ? (3000) : (5000)) |
| 66 | crit: $this > (($status == $CRITICAL) ? (5000) : (10000)) |
| 67 | delay: down 5m multiplier 1.5 max 1h |
| 68 | summary: Cognitive Services latency on ${label:resource_name} |
| 69 | info: Average API latency of Cognitive Services resource ${label:resource_name} \ |
| 70 | in ${label:resource_group} (${label:region}) |
| 71 | to: sysadmin |
| 72 | |
| 73 | # --- Cognitive Services - Rate Limiting --- |
| 74 | |
| 75 | template: am_cognitive_services_rate_limit |
| 76 | on: azure_monitor.cognitive_services.rate_limit |
| 77 | class: Workload |
| 78 | type: Other |
| 79 | component: Cognitive Services |
| 80 | lookup: average -5m unaligned of rate_limit |
| 81 | units: requests/s |
| 82 | every: 1m |
| 83 | warn: $this > (($status >= $WARNING) ? (1) : (5)) |
| 84 | crit: $this > (($status == $CRITICAL) ? (5) : (20)) |
| 85 | delay: down 5m multiplier 1.5 max 1h |
| 86 | summary: Cognitive Services rate limiting on ${label:resource_name} |
| 87 | info: Rate of throttled requests due to rate limiting on Cognitive Services resource ${label:resource_name} \ |
| 88 | in ${label:resource_group} (${label:region}). \ |
| 89 | Indicates the service is hitting its API call quota |
| 90 | to: sysadmin |
| 91 | |
| 92 | # --- Cognitive Services - Blocked Calls --- |
| 93 | |
| 94 | template: am_cognitive_services_blocked_calls |
| 95 | on: azure_monitor.cognitive_services.calls |
| 96 | class: Errors |
| 97 | type: Other |
| 98 | component: Cognitive Services |
| 99 | lookup: average -5m unaligned of blocked |
| 100 | units: calls/s |
| 101 | every: 1m |
| 102 | warn: $this > (($status >= $WARNING) ? (0) : (5)) |
| 103 | crit: $this > (($status == $CRITICAL) ? (5) : (20)) |
| 104 | delay: down 5m multiplier 1.5 max 1h |
| 105 | summary: Cognitive Services blocked calls on ${label:resource_name} |
| 106 | info: Rate of blocked API calls on Cognitive Services resource ${label:resource_name} \ |
| 107 | in ${label:resource_group} (${label:region}). \ |
| 108 | Blocked calls indicate authorization or policy violations |
| 109 | to: sysadmin |
| 110 | |
| 111 | # --- Azure OpenAI - Availability --- |
| 112 | |
| 113 | template: am_cognitive_services_openai_availability |
| 114 | on: azure_monitor.cognitive_services.openai_availability |
| 115 | class: Availability |
| 116 | type: Other |
| 117 | component: Cognitive Services |
| 118 | lookup: average -5m unaligned of availability |
| 119 | units: percentage |
| 120 | every: 1m |
| 121 | warn: $this != nan AND $this < (($status >= $WARNING) ? (99.9) : (99)) |
| 122 | crit: $this != nan AND $this < (($status == $CRITICAL) ? (99) : (90)) |
| 123 | delay: down 5m multiplier 1.5 max 1h |
| 124 | summary: Azure OpenAI availability on ${label:resource_name} |
| 125 | info: Availability rate of Azure OpenAI service on ${label:resource_name} \ |
| 126 | in ${label:resource_group} (${label:region}) |
| 127 | to: sysadmin |
| 128 | |
| 129 | # --- Azure OpenAI - Latency --- |
| 130 | |
| 131 | template: am_cognitive_services_openai_time_to_response |
| 132 | on: azure_monitor.cognitive_services.openai_latency |
| 133 | class: Latency |
| 134 | type: Other |
| 135 | component: Cognitive Services |
| 136 | lookup: average -5m unaligned of time_to_response |
| 137 | units: milliseconds |
| 138 | every: 1m |
| 139 | warn: $this != nan AND $this > (($status >= $WARNING) ? (5000) : (10000)) |
| 140 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (10000) : (30000)) |
| 141 | delay: down 5m multiplier 1.5 max 1h |
| 142 | summary: Azure OpenAI time to response on ${label:resource_name} |
| 143 | info: Average time to response for Azure OpenAI requests on ${label:resource_name} \ |
| 144 | in ${label:resource_group} (${label:region}) |
| 145 | to: sysadmin |
| 146 | |
| 147 | template: am_cognitive_services_openai_time_to_first_token |
| 148 | on: azure_monitor.cognitive_services.openai_latency |
| 149 | class: Latency |
| 150 | type: Other |
| 151 | component: Cognitive Services |
| 152 | lookup: average -5m unaligned of time_to_first_token |
| 153 | units: milliseconds |
| 154 | every: 1m |
| 155 | warn: $this != nan AND $this > (($status >= $WARNING) ? (3000) : (5000)) |
| 156 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (5000) : (10000)) |
| 157 | delay: down 5m multiplier 1.5 max 1h |
| 158 | summary: Azure OpenAI time to first token on ${label:resource_name} |
| 159 | info: Average normalized time to first token for Azure OpenAI requests on ${label:resource_name} \ |
| 160 | in ${label:resource_group} (${label:region}) |
| 161 | to: sysadmin |
| 162 | |
| 163 | # --- Azure OpenAI - Provisioned Utilization --- |
| 164 | |
| 165 | template: am_cognitive_services_openai_provisioned_utilization |
| 166 | on: azure_monitor.cognitive_services.openai_provisioned_utilization |
| 167 | class: Utilization |
| 168 | type: Other |
| 169 | component: Cognitive Services |
| 170 | lookup: average -5m unaligned of utilization |
| 171 | units: percentage |
| 172 | every: 1m |
| 173 | warn: $this != nan AND $this > (($status >= $WARNING) ? (70) : (80)) |
| 174 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (80) : (90)) |
| 175 | delay: down 5m multiplier 1.5 max 1h |
| 176 | summary: Azure OpenAI provisioned utilization on ${label:resource_name} |
| 177 | info: Provisioned-managed utilization of Azure OpenAI deployment on ${label:resource_name} \ |
| 178 | in ${label:resource_group} (${label:region}). \ |
| 179 | High utilization means requests may be throttled or rejected |
| 180 | to: sysadmin |
| 181 | |
| 182 | # --- Models - Availability --- |
| 183 | |
| 184 | template: am_cognitive_services_model_availability |
| 185 | on: azure_monitor.cognitive_services.model_availability |
| 186 | class: Availability |
| 187 | type: Other |
| 188 | component: Cognitive Services |
| 189 | lookup: average -5m unaligned of availability |
| 190 | units: percentage |
| 191 | every: 1m |
| 192 | warn: $this != nan AND $this < (($status >= $WARNING) ? (99.9) : (99)) |
| 193 | crit: $this != nan AND $this < (($status == $CRITICAL) ? (99) : (90)) |
| 194 | delay: down 5m multiplier 1.5 max 1h |
| 195 | summary: Model availability on ${label:resource_name} |
| 196 | info: Model availability rate on Cognitive Services resource ${label:resource_name} \ |
| 197 | in ${label:resource_group} (${label:region}) |
| 198 | to: sysadmin |
| 199 | |
| 200 | # --- Models - Latency --- |
| 201 | |
| 202 | template: am_cognitive_services_model_time_to_response |
| 203 | on: azure_monitor.cognitive_services.model_latency |
| 204 | class: Latency |
| 205 | type: Other |
| 206 | component: Cognitive Services |
| 207 | lookup: average -5m unaligned of time_to_response |
| 208 | units: milliseconds |
| 209 | every: 1m |
| 210 | warn: $this != nan AND $this > (($status >= $WARNING) ? (5000) : (10000)) |
| 211 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (10000) : (30000)) |
| 212 | delay: down 5m multiplier 1.5 max 1h |
| 213 | summary: Model time to response on ${label:resource_name} |
| 214 | info: Average time to response for model requests on Cognitive Services resource ${label:resource_name} \ |
| 215 | in ${label:resource_group} (${label:region}) |
| 216 | to: sysadmin |
| 217 | |
| 218 | template: am_cognitive_services_model_time_to_first_token |
| 219 | on: azure_monitor.cognitive_services.model_latency |
| 220 | class: Latency |
| 221 | type: Other |
| 222 | component: Cognitive Services |
| 223 | lookup: average -5m unaligned of time_to_first_token |
| 224 | units: milliseconds |
| 225 | every: 1m |
| 226 | warn: $this != nan AND $this > (($status >= $WARNING) ? (3000) : (5000)) |
| 227 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (5000) : (10000)) |
| 228 | delay: down 5m multiplier 1.5 max 1h |
| 229 | summary: Model time to first token on ${label:resource_name} |
| 230 | info: Average normalized time to first token for model requests on Cognitive Services resource ${label:resource_name} \ |
| 231 | in ${label:resource_group} (${label:region}) |
| 232 | to: sysadmin |
| 233 | |
| 234 | # --- Models - Provisioned Utilization --- |
| 235 | |
| 236 | template: am_cognitive_services_model_provisioned_utilization |
| 237 | on: azure_monitor.cognitive_services.model_provisioned_utilization |
| 238 | class: Utilization |
| 239 | type: Other |
| 240 | component: Cognitive Services |
| 241 | lookup: average -5m unaligned of utilization |
| 242 | units: percentage |
| 243 | every: 1m |
| 244 | warn: $this != nan AND $this > (($status >= $WARNING) ? (70) : (80)) |
| 245 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (80) : (90)) |
| 246 | delay: down 5m multiplier 1.5 max 1h |
| 247 | summary: Model provisioned utilization on ${label:resource_name} |
| 248 | info: Provisioned utilization of model deployment on Cognitive Services resource ${label:resource_name} \ |
| 249 | in ${label:resource_group} (${label:region}). \ |
| 250 | High utilization means requests may be throttled or rejected |
| 251 | to: sysadmin |
| 252 | |
| 253 | # --- Content Safety - Harmful Requests --- |
| 254 | |
| 255 | template: am_cognitive_services_harmful_requests |
| 256 | on: azure_monitor.cognitive_services.content_safety_requests |
| 257 | class: Errors |
| 258 | type: Other |
| 259 | component: Cognitive Services |
| 260 | lookup: average -5m unaligned of harmful |
| 261 | units: requests/s |
| 262 | every: 1m |
| 263 | warn: $this != nan AND $this > (($status >= $WARNING) ? (1) : (5)) |
| 264 | crit: $this != nan AND $this > (($status == $CRITICAL) ? (5) : (20)) |
| 265 | delay: down 5m multiplier 1.5 max 1h |
| 266 | summary: Harmful content requests on ${label:resource_name} |
| 267 | info: Rate of requests flagged as harmful by content safety on Cognitive Services resource ${label:resource_name} \ |
| 268 | in ${label:resource_group} (${label:region}). \ |
| 269 | High rates may indicate abuse or prompt injection attempts |
| 270 | to: sysadmin |
| 271 | |
| 272 | template: am_cognitive_services_blocked_content_requests |
| 273 | on: azure_monitor.cognitive_services.content_safety_requests |
| 274 | class: Errors |
| 275 | type: Other |
| 276 | component: Cognitive Services |
| 277 | lookup: average -5m unaligned of blocked |
| 278 | units: requests/s |
| 279 | every: 1m |
| 280 | warn: $this != nan AND $this > (($status >= $WARNING) ? (1) : (10)) |
| 281 | delay: down 5m multiplier 1.5 max 1h |
| 282 | summary: Blocked content requests on ${label:resource_name} |
| 283 | info: Rate of requests rejected by content safety on Cognitive Services resource ${label:resource_name} \ |
| 284 | in ${label:resource_group} (${label:region}) |
| 285 | to: sysadmin |
| 286 | |
| 287 | # --- Content Safety - Abusive Users --- |
| 288 | |
| 289 | template: am_cognitive_services_abusive_users |
| 290 | on: azure_monitor.cognitive_services.content_safety_abusive_users |
| 291 | class: Errors |
| 292 | type: Other |
| 293 | component: Cognitive Services |
| 294 | lookup: average -5m unaligned of abusive_users |
| 295 | units: users/s |
| 296 | every: 1m |
| 297 | warn: $this != nan AND $this > (($status >= $WARNING) ? (0) : (1)) |
| 298 | delay: down 5m multiplier 1.5 max 1h |
| 299 | summary: Abusive users detected on ${label:resource_name} |
| 300 | info: Rate of potentially abusive users detected on Cognitive Services resource ${label:resource_name} \ |
| 301 | in ${label:resource_group} (${label:region}) |
| 302 | to: sysadmin |