[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-nvidia-dynamo-troubleshoot":3,"mdc-fiygp1-key":33,"related-org-nvidia-dynamo-troubleshoot":1000,"related-repo-nvidia-dynamo-troubleshoot":1156},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":22,"repoUrl":23,"updatedAt":24,"license":25,"forks":26,"topics":27,"repo":28,"sourceUrl":31,"mdContent":32},"dynamo-troubleshoot","troubleshoot Dynamo deployments","Diagnose failed or unhealthy Dynamo deployments. Use when pods, model-cache jobs, PVCs, workers, frontend\u002Frouter health, endpoints, or benchmark jobs fail; use recipe-runner\u002Frouter-starter before this for normal bring-up.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},"nvidia","NVIDIA","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fnvidia.png",[12,16,19],{"name":13,"slug":14,"type":15},"Deployment","deployment","tag",{"name":17,"slug":18,"type":15},"Kubernetes","kubernetes",{"name":20,"slug":21,"type":15},"Debugging","debugging",2473,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills","2026-07-14T05:28:29.579113","Apache-2.0",281,[],{"repoUrl":23,"stars":22,"forks":26,"topics":29,"description":30},[],"AI agent skills published by NVIDIA","https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Fdynamo-troubleshoot","---\nname: dynamo-troubleshoot\ndescription: Diagnose failed or unhealthy Dynamo deployments. Use when pods, model-cache jobs, PVCs, workers, frontend\u002Frouter health, endpoints, or benchmark jobs fail; use recipe-runner\u002Frouter-starter before this for normal bring-up.\nlicense: Apache-2.0\nmetadata:\n  author: Dan Gil \u003Cdagil@nvidia.com>\n  tags:\n    - dynamo\n    - kubernetes\n    - troubleshooting\n    - day-2\n---\n\n# Dynamo Troubleshoot\n\n\u003C!--\nSPDX-FileCopyrightText: Copyright (c) 2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.\nSPDX-License-Identifier: CC-BY-4.0\n-->\n\n## Purpose\n\nTurn a Dynamo failure into a clear problem class, strongest signal, and next\naction. Start with read-only evidence, avoid secrets, and fix one layer at a\ntime.\n\n## Prerequisites\n\n- Python 3.10+ on the operator machine.\n- `kubectl` configured with read access to the target namespace.\n- Permission to read pods, events, jobs, PVCs, and `DynamoGraphDeployment` resources (NOT secrets).\n- Network reachability to the cluster API server.\n\n## Instructions\n\n### 1. Collect A Read-Only Bundle\n\nRun:\n\n```bash\npython3 scripts\u002Fcollect_dynamo_debug_bundle.py \\\n  --namespace \"${NAMESPACE}\"\n```\n\nIf the user names a deployment, include it:\n\n```bash\npython3 scripts\u002Fcollect_dynamo_debug_bundle.py \\\n  --namespace \"${NAMESPACE}\" \\\n  --deployment-name \u003Cdeployment-name>\n```\n\nDo not collect Kubernetes secrets. Do not print Hugging Face tokens.\n\n### 2. Classify The Failure\n\nUse `references\u002Ffailure-decision-tree.md` and classify into one primary bucket:\n\n- cluster\u002Fplatform\n- namespace\u002Fsecret\n- model cache\u002FPVC\u002Fdownload\n- image pull\u002Fruntime image\n- GPU scheduling\u002Fresources\n- operator\u002FDynamoGraphDeployment reconciliation\n- frontend\u002Frouter\n- worker\u002Fbackend\n- endpoint\u002FAPI\n- benchmark\u002Fperf job\n\n### 3. Debug Top Down\n\nCheck in this order:\n\n1. namespace, storage class, GPU nodes, and HF secret existence\n2. PVC and model-download job\n3. `DynamoGraphDeployment` status and events\n4. pod status, `describe pod`, and container logs\n5. frontend service and port-forward\n6. `\u002Fv1\u002Fmodels`\n7. `\u002Fv1\u002Fchat\u002Fcompletions`\n8. benchmark job only after endpoint smoke test passes\n\n### 4. Fix One Layer At A Time\n\nPrefer the smallest reversible change:\n\n- create missing namespace or HF secret\n- patch `storageClassName`\n- patch image tag or image pull secret\n- reduce GPU request only if the recipe can still be valid\n- switch KV router to approximate mode only if workers do not publish events\n- restart failed jobs after fixing the underlying config\n\nAfter each fix, rerun the relevant readiness check before moving deeper.\n\n## Available Scripts\n\n| Script | Purpose | Arguments |\n|---|---|---|\n| `scripts\u002Fcollect_dynamo_debug_bundle.py` | Collect a read-only debug bundle (pods, events, jobs, PVCs, CR status) | `--namespace`, `--deployment-name`, `--output-dir` |\n\nInvoke via the agentskills.io `run_script()` protocol:\n\n```python\nrun_script(\"scripts\u002Fcollect_dynamo_debug_bundle.py\", args=[\"--namespace\", \"dynamo-demo\"])\n```\n\n## Examples\n\nCollect everything in a namespace for triage:\n\n```bash\npython3 scripts\u002Fcollect_dynamo_debug_bundle.py --namespace dynamo-demo\n```\n\nScope to a single failing deployment:\n\n```bash\npython3 scripts\u002Fcollect_dynamo_debug_bundle.py \\\n  --namespace dynamo-demo \\\n  --deployment-name qwen-vllm-disagg\n```\n\nEquivalent through the agent protocol:\n\n```python\nrun_script(\"scripts\u002Fcollect_dynamo_debug_bundle.py\", args=[\"--namespace\", \"dynamo-demo\", \"--deployment-name\", \"qwen-vllm-disagg\"])\n```\n\n## Output Contract\n\nReturn:\n\n- problem class\n- evidence checked\n- strongest signal\n- likely cause\n- exact next command or patch\n- what was ruled out\n- whether it is safe to continue deployment or benchmarking\n\n## Limitations\n\n- Read-only. Never mutates the cluster; remediation commands are returned, not executed.\n- Will not collect secrets or print Hugging Face tokens; some failure modes (auth) may need user-side inspection.\n- Bundle size grows with deployment size; on very large namespaces, scope with `--deployment-name`.\n- Does not validate disagg transport — use `dynamo-interconnect-check` for that.\n\n## Troubleshooting\n\n| Symptom | Likely cause | Next step |\n|---|---|---|\n| `kubectl` returns Forbidden on events\u002Fpods | Service account lacks read RBAC | Ask operator for read-only role binding on the namespace |\n| Bundle missing `DynamoGraphDeployment` status | Operator not installed or different namespace | Verify `dynamo-platform` operator is installed and watching the namespace |\n| Model-download job in `Pending` | PVC unbound or HF secret missing | Fix PVC binding or create the named HF secret, then rerun the job |\n| Worker pods `CrashLoopBackOff` | Image\u002Fruntime mismatch or GPU not available | Inspect container logs; check `nvidia.com\u002Fgpu` allocatable on nodes |\n\n## Benchmark\n\nSee `BENCHMARK.md` for the NVCARPS-EVAL performance report (auto-generated by the NVSkills CI pipeline). To refresh, re-run `\u002Fnvskills-ci` on an upstream PR touching this skill.\n\n## References\n\n- Read `references\u002Ffailure-decision-tree.md` for bucket-specific checks.\n- Use `scripts\u002Fcollect_dynamo_debug_bundle.py` for read-only bundle collection.\n",{"data":34,"body":41},{"name":4,"description":6,"license":25,"metadata":35},{"author":36,"tags":37},"Dan Gil \u003Cdagil@nvidia.com>",[38,18,39,40],"dynamo","troubleshooting","day-2",{"type":42,"children":43},"root",[44,52,59,65,71,111,117,124,129,189,194,269,274,280,293,346,352,357,422,428,433,472,477,483,556,569,585,591,596,624,629,679,684,698,704,709,747,753,791,796,935,941,962,968,994],{"type":45,"tag":46,"props":47,"children":48},"element","h1",{"id":4},[49],{"type":50,"value":51},"text","Dynamo Troubleshoot",{"type":45,"tag":53,"props":54,"children":56},"h2",{"id":55},"purpose",[57],{"type":50,"value":58},"Purpose",{"type":45,"tag":60,"props":61,"children":62},"p",{},[63],{"type":50,"value":64},"Turn a Dynamo failure into a clear problem class, strongest signal, and next\naction. Start with read-only evidence, avoid secrets, and fix one layer at a\ntime.",{"type":45,"tag":53,"props":66,"children":68},{"id":67},"prerequisites",[69],{"type":50,"value":70},"Prerequisites",{"type":45,"tag":72,"props":73,"children":74},"ul",{},[75,81,93,106],{"type":45,"tag":76,"props":77,"children":78},"li",{},[79],{"type":50,"value":80},"Python 3.10+ on the operator machine.",{"type":45,"tag":76,"props":82,"children":83},{},[84,91],{"type":45,"tag":85,"props":86,"children":88},"code",{"className":87},[],[89],{"type":50,"value":90},"kubectl",{"type":50,"value":92}," configured with read access to the target namespace.",{"type":45,"tag":76,"props":94,"children":95},{},[96,98,104],{"type":50,"value":97},"Permission to read pods, events, jobs, PVCs, and ",{"type":45,"tag":85,"props":99,"children":101},{"className":100},[],[102],{"type":50,"value":103},"DynamoGraphDeployment",{"type":50,"value":105}," resources (NOT secrets).",{"type":45,"tag":76,"props":107,"children":108},{},[109],{"type":50,"value":110},"Network reachability to the cluster API server.",{"type":45,"tag":53,"props":112,"children":114},{"id":113},"instructions",[115],{"type":50,"value":116},"Instructions",{"type":45,"tag":118,"props":119,"children":121},"h3",{"id":120},"_1-collect-a-read-only-bundle",[122],{"type":50,"value":123},"1. Collect A Read-Only Bundle",{"type":45,"tag":60,"props":125,"children":126},{},[127],{"type":50,"value":128},"Run:",{"type":45,"tag":130,"props":131,"children":136},"pre",{"className":132,"code":133,"language":134,"meta":135,"style":135},"language-bash shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","python3 scripts\u002Fcollect_dynamo_debug_bundle.py \\\n  --namespace \"${NAMESPACE}\"\n","bash","",[137],{"type":45,"tag":85,"props":138,"children":139},{"__ignoreMap":135},[140,164],{"type":45,"tag":141,"props":142,"children":145},"span",{"class":143,"line":144},"line",1,[146,152,158],{"type":45,"tag":141,"props":147,"children":149},{"style":148},"--shiki-light:#E2931D;--shiki-default:#FFCB6B;--shiki-dark:#FFCB6B",[150],{"type":50,"value":151},"python3",{"type":45,"tag":141,"props":153,"children":155},{"style":154},"--shiki-light:#91B859;--shiki-default:#C3E88D;--shiki-dark:#C3E88D",[156],{"type":50,"value":157}," scripts\u002Fcollect_dynamo_debug_bundle.py",{"type":45,"tag":141,"props":159,"children":161},{"style":160},"--shiki-light:#90A4AE;--shiki-default:#EEFFFF;--shiki-dark:#BABED8",[162],{"type":50,"value":163}," \\\n",{"type":45,"tag":141,"props":165,"children":167},{"class":143,"line":166},2,[168,173,179,184],{"type":45,"tag":141,"props":169,"children":170},{"style":154},[171],{"type":50,"value":172},"  --namespace",{"type":45,"tag":141,"props":174,"children":176},{"style":175},"--shiki-light:#39ADB5;--shiki-default:#89DDFF;--shiki-dark:#89DDFF",[177],{"type":50,"value":178}," \"${",{"type":45,"tag":141,"props":180,"children":181},{"style":160},[182],{"type":50,"value":183},"NAMESPACE",{"type":45,"tag":141,"props":185,"children":186},{"style":175},[187],{"type":50,"value":188},"}\"\n",{"type":45,"tag":60,"props":190,"children":191},{},[192],{"type":50,"value":193},"If the user names a deployment, include it:",{"type":45,"tag":130,"props":195,"children":197},{"className":132,"code":196,"language":134,"meta":135,"style":135},"python3 scripts\u002Fcollect_dynamo_debug_bundle.py \\\n  --namespace \"${NAMESPACE}\" \\\n  --deployment-name \u003Cdeployment-name>\n",[198],{"type":45,"tag":85,"props":199,"children":200},{"__ignoreMap":135},[201,216,240],{"type":45,"tag":141,"props":202,"children":203},{"class":143,"line":144},[204,208,212],{"type":45,"tag":141,"props":205,"children":206},{"style":148},[207],{"type":50,"value":151},{"type":45,"tag":141,"props":209,"children":210},{"style":154},[211],{"type":50,"value":157},{"type":45,"tag":141,"props":213,"children":214},{"style":160},[215],{"type":50,"value":163},{"type":45,"tag":141,"props":217,"children":218},{"class":143,"line":166},[219,223,227,231,236],{"type":45,"tag":141,"props":220,"children":221},{"style":154},[222],{"type":50,"value":172},{"type":45,"tag":141,"props":224,"children":225},{"style":175},[226],{"type":50,"value":178},{"type":45,"tag":141,"props":228,"children":229},{"style":160},[230],{"type":50,"value":183},{"type":45,"tag":141,"props":232,"children":233},{"style":175},[234],{"type":50,"value":235},"}\"",{"type":45,"tag":141,"props":237,"children":238},{"style":160},[239],{"type":50,"value":163},{"type":45,"tag":141,"props":241,"children":243},{"class":143,"line":242},3,[244,249,254,259,264],{"type":45,"tag":141,"props":245,"children":246},{"style":154},[247],{"type":50,"value":248},"  --deployment-name",{"type":45,"tag":141,"props":250,"children":251},{"style":175},[252],{"type":50,"value":253}," \u003C",{"type":45,"tag":141,"props":255,"children":256},{"style":154},[257],{"type":50,"value":258},"deployment-nam",{"type":45,"tag":141,"props":260,"children":261},{"style":160},[262],{"type":50,"value":263},"e",{"type":45,"tag":141,"props":265,"children":266},{"style":175},[267],{"type":50,"value":268},">\n",{"type":45,"tag":60,"props":270,"children":271},{},[272],{"type":50,"value":273},"Do not collect Kubernetes secrets. Do not print Hugging Face tokens.",{"type":45,"tag":118,"props":275,"children":277},{"id":276},"_2-classify-the-failure",[278],{"type":50,"value":279},"2. Classify The Failure",{"type":45,"tag":60,"props":281,"children":282},{},[283,285,291],{"type":50,"value":284},"Use ",{"type":45,"tag":85,"props":286,"children":288},{"className":287},[],[289],{"type":50,"value":290},"references\u002Ffailure-decision-tree.md",{"type":50,"value":292}," and classify into one primary bucket:",{"type":45,"tag":72,"props":294,"children":295},{},[296,301,306,311,316,321,326,331,336,341],{"type":45,"tag":76,"props":297,"children":298},{},[299],{"type":50,"value":300},"cluster\u002Fplatform",{"type":45,"tag":76,"props":302,"children":303},{},[304],{"type":50,"value":305},"namespace\u002Fsecret",{"type":45,"tag":76,"props":307,"children":308},{},[309],{"type":50,"value":310},"model cache\u002FPVC\u002Fdownload",{"type":45,"tag":76,"props":312,"children":313},{},[314],{"type":50,"value":315},"image pull\u002Fruntime image",{"type":45,"tag":76,"props":317,"children":318},{},[319],{"type":50,"value":320},"GPU scheduling\u002Fresources",{"type":45,"tag":76,"props":322,"children":323},{},[324],{"type":50,"value":325},"operator\u002FDynamoGraphDeployment reconciliation",{"type":45,"tag":76,"props":327,"children":328},{},[329],{"type":50,"value":330},"frontend\u002Frouter",{"type":45,"tag":76,"props":332,"children":333},{},[334],{"type":50,"value":335},"worker\u002Fbackend",{"type":45,"tag":76,"props":337,"children":338},{},[339],{"type":50,"value":340},"endpoint\u002FAPI",{"type":45,"tag":76,"props":342,"children":343},{},[344],{"type":50,"value":345},"benchmark\u002Fperf job",{"type":45,"tag":118,"props":347,"children":349},{"id":348},"_3-debug-top-down",[350],{"type":50,"value":351},"3. Debug Top Down",{"type":45,"tag":60,"props":353,"children":354},{},[355],{"type":50,"value":356},"Check in this order:",{"type":45,"tag":358,"props":359,"children":360},"ol",{},[361,366,371,381,394,399,408,417],{"type":45,"tag":76,"props":362,"children":363},{},[364],{"type":50,"value":365},"namespace, storage class, GPU nodes, and HF secret existence",{"type":45,"tag":76,"props":367,"children":368},{},[369],{"type":50,"value":370},"PVC and model-download job",{"type":45,"tag":76,"props":372,"children":373},{},[374,379],{"type":45,"tag":85,"props":375,"children":377},{"className":376},[],[378],{"type":50,"value":103},{"type":50,"value":380}," status and events",{"type":45,"tag":76,"props":382,"children":383},{},[384,386,392],{"type":50,"value":385},"pod status, ",{"type":45,"tag":85,"props":387,"children":389},{"className":388},[],[390],{"type":50,"value":391},"describe pod",{"type":50,"value":393},", and container logs",{"type":45,"tag":76,"props":395,"children":396},{},[397],{"type":50,"value":398},"frontend service and port-forward",{"type":45,"tag":76,"props":400,"children":401},{},[402],{"type":45,"tag":85,"props":403,"children":405},{"className":404},[],[406],{"type":50,"value":407},"\u002Fv1\u002Fmodels",{"type":45,"tag":76,"props":409,"children":410},{},[411],{"type":45,"tag":85,"props":412,"children":414},{"className":413},[],[415],{"type":50,"value":416},"\u002Fv1\u002Fchat\u002Fcompletions",{"type":45,"tag":76,"props":418,"children":419},{},[420],{"type":50,"value":421},"benchmark job only after endpoint smoke test passes",{"type":45,"tag":118,"props":423,"children":425},{"id":424},"_4-fix-one-layer-at-a-time",[426],{"type":50,"value":427},"4. Fix One Layer At A Time",{"type":45,"tag":60,"props":429,"children":430},{},[431],{"type":50,"value":432},"Prefer the smallest reversible change:",{"type":45,"tag":72,"props":434,"children":435},{},[436,441,452,457,462,467],{"type":45,"tag":76,"props":437,"children":438},{},[439],{"type":50,"value":440},"create missing namespace or HF secret",{"type":45,"tag":76,"props":442,"children":443},{},[444,446],{"type":50,"value":445},"patch ",{"type":45,"tag":85,"props":447,"children":449},{"className":448},[],[450],{"type":50,"value":451},"storageClassName",{"type":45,"tag":76,"props":453,"children":454},{},[455],{"type":50,"value":456},"patch image tag or image pull secret",{"type":45,"tag":76,"props":458,"children":459},{},[460],{"type":50,"value":461},"reduce GPU request only if the recipe can still be valid",{"type":45,"tag":76,"props":463,"children":464},{},[465],{"type":50,"value":466},"switch KV router to approximate mode only if workers do not publish events",{"type":45,"tag":76,"props":468,"children":469},{},[470],{"type":50,"value":471},"restart failed jobs after fixing the underlying config",{"type":45,"tag":60,"props":473,"children":474},{},[475],{"type":50,"value":476},"After each fix, rerun the relevant readiness check before moving deeper.",{"type":45,"tag":53,"props":478,"children":480},{"id":479},"available-scripts",[481],{"type":50,"value":482},"Available Scripts",{"type":45,"tag":484,"props":485,"children":486},"table",{},[487,510],{"type":45,"tag":488,"props":489,"children":490},"thead",{},[491],{"type":45,"tag":492,"props":493,"children":494},"tr",{},[495,501,505],{"type":45,"tag":496,"props":497,"children":498},"th",{},[499],{"type":50,"value":500},"Script",{"type":45,"tag":496,"props":502,"children":503},{},[504],{"type":50,"value":58},{"type":45,"tag":496,"props":506,"children":507},{},[508],{"type":50,"value":509},"Arguments",{"type":45,"tag":511,"props":512,"children":513},"tbody",{},[514],{"type":45,"tag":492,"props":515,"children":516},{},[517,527,532],{"type":45,"tag":518,"props":519,"children":520},"td",{},[521],{"type":45,"tag":85,"props":522,"children":524},{"className":523},[],[525],{"type":50,"value":526},"scripts\u002Fcollect_dynamo_debug_bundle.py",{"type":45,"tag":518,"props":528,"children":529},{},[530],{"type":50,"value":531},"Collect a read-only debug bundle (pods, events, jobs, PVCs, CR status)",{"type":45,"tag":518,"props":533,"children":534},{},[535,541,543,549,550],{"type":45,"tag":85,"props":536,"children":538},{"className":537},[],[539],{"type":50,"value":540},"--namespace",{"type":50,"value":542},", ",{"type":45,"tag":85,"props":544,"children":546},{"className":545},[],[547],{"type":50,"value":548},"--deployment-name",{"type":50,"value":542},{"type":45,"tag":85,"props":551,"children":553},{"className":552},[],[554],{"type":50,"value":555},"--output-dir",{"type":45,"tag":60,"props":557,"children":558},{},[559,561,567],{"type":50,"value":560},"Invoke via the agentskills.io ",{"type":45,"tag":85,"props":562,"children":564},{"className":563},[],[565],{"type":50,"value":566},"run_script()",{"type":50,"value":568}," protocol:",{"type":45,"tag":130,"props":570,"children":574},{"className":571,"code":572,"language":573,"meta":135,"style":135},"language-python shiki shiki-themes material-theme-lighter material-theme material-theme-palenight","run_script(\"scripts\u002Fcollect_dynamo_debug_bundle.py\", args=[\"--namespace\", \"dynamo-demo\"])\n","python",[575],{"type":45,"tag":85,"props":576,"children":577},{"__ignoreMap":135},[578],{"type":45,"tag":141,"props":579,"children":580},{"class":143,"line":144},[581],{"type":45,"tag":141,"props":582,"children":583},{},[584],{"type":50,"value":572},{"type":45,"tag":53,"props":586,"children":588},{"id":587},"examples",[589],{"type":50,"value":590},"Examples",{"type":45,"tag":60,"props":592,"children":593},{},[594],{"type":50,"value":595},"Collect everything in a namespace for triage:",{"type":45,"tag":130,"props":597,"children":599},{"className":132,"code":598,"language":134,"meta":135,"style":135},"python3 scripts\u002Fcollect_dynamo_debug_bundle.py --namespace dynamo-demo\n",[600],{"type":45,"tag":85,"props":601,"children":602},{"__ignoreMap":135},[603],{"type":45,"tag":141,"props":604,"children":605},{"class":143,"line":144},[606,610,614,619],{"type":45,"tag":141,"props":607,"children":608},{"style":148},[609],{"type":50,"value":151},{"type":45,"tag":141,"props":611,"children":612},{"style":154},[613],{"type":50,"value":157},{"type":45,"tag":141,"props":615,"children":616},{"style":154},[617],{"type":50,"value":618}," --namespace",{"type":45,"tag":141,"props":620,"children":621},{"style":154},[622],{"type":50,"value":623}," dynamo-demo\n",{"type":45,"tag":60,"props":625,"children":626},{},[627],{"type":50,"value":628},"Scope to a single failing deployment:",{"type":45,"tag":130,"props":630,"children":632},{"className":132,"code":631,"language":134,"meta":135,"style":135},"python3 scripts\u002Fcollect_dynamo_debug_bundle.py \\\n  --namespace dynamo-demo \\\n  --deployment-name qwen-vllm-disagg\n",[633],{"type":45,"tag":85,"props":634,"children":635},{"__ignoreMap":135},[636,651,667],{"type":45,"tag":141,"props":637,"children":638},{"class":143,"line":144},[639,643,647],{"type":45,"tag":141,"props":640,"children":641},{"style":148},[642],{"type":50,"value":151},{"type":45,"tag":141,"props":644,"children":645},{"style":154},[646],{"type":50,"value":157},{"type":45,"tag":141,"props":648,"children":649},{"style":160},[650],{"type":50,"value":163},{"type":45,"tag":141,"props":652,"children":653},{"class":143,"line":166},[654,658,663],{"type":45,"tag":141,"props":655,"children":656},{"style":154},[657],{"type":50,"value":172},{"type":45,"tag":141,"props":659,"children":660},{"style":154},[661],{"type":50,"value":662}," dynamo-demo",{"type":45,"tag":141,"props":664,"children":665},{"style":160},[666],{"type":50,"value":163},{"type":45,"tag":141,"props":668,"children":669},{"class":143,"line":242},[670,674],{"type":45,"tag":141,"props":671,"children":672},{"style":154},[673],{"type":50,"value":248},{"type":45,"tag":141,"props":675,"children":676},{"style":154},[677],{"type":50,"value":678}," qwen-vllm-disagg\n",{"type":45,"tag":60,"props":680,"children":681},{},[682],{"type":50,"value":683},"Equivalent through the agent protocol:",{"type":45,"tag":130,"props":685,"children":687},{"className":571,"code":686,"language":573,"meta":135,"style":135},"run_script(\"scripts\u002Fcollect_dynamo_debug_bundle.py\", args=[\"--namespace\", \"dynamo-demo\", \"--deployment-name\", \"qwen-vllm-disagg\"])\n",[688],{"type":45,"tag":85,"props":689,"children":690},{"__ignoreMap":135},[691],{"type":45,"tag":141,"props":692,"children":693},{"class":143,"line":144},[694],{"type":45,"tag":141,"props":695,"children":696},{},[697],{"type":50,"value":686},{"type":45,"tag":53,"props":699,"children":701},{"id":700},"output-contract",[702],{"type":50,"value":703},"Output Contract",{"type":45,"tag":60,"props":705,"children":706},{},[707],{"type":50,"value":708},"Return:",{"type":45,"tag":72,"props":710,"children":711},{},[712,717,722,727,732,737,742],{"type":45,"tag":76,"props":713,"children":714},{},[715],{"type":50,"value":716},"problem class",{"type":45,"tag":76,"props":718,"children":719},{},[720],{"type":50,"value":721},"evidence checked",{"type":45,"tag":76,"props":723,"children":724},{},[725],{"type":50,"value":726},"strongest signal",{"type":45,"tag":76,"props":728,"children":729},{},[730],{"type":50,"value":731},"likely cause",{"type":45,"tag":76,"props":733,"children":734},{},[735],{"type":50,"value":736},"exact next command or patch",{"type":45,"tag":76,"props":738,"children":739},{},[740],{"type":50,"value":741},"what was ruled out",{"type":45,"tag":76,"props":743,"children":744},{},[745],{"type":50,"value":746},"whether it is safe to continue deployment or benchmarking",{"type":45,"tag":53,"props":748,"children":750},{"id":749},"limitations",[751],{"type":50,"value":752},"Limitations",{"type":45,"tag":72,"props":754,"children":755},{},[756,761,766,778],{"type":45,"tag":76,"props":757,"children":758},{},[759],{"type":50,"value":760},"Read-only. Never mutates the cluster; remediation commands are returned, not executed.",{"type":45,"tag":76,"props":762,"children":763},{},[764],{"type":50,"value":765},"Will not collect secrets or print Hugging Face tokens; some failure modes (auth) may need user-side inspection.",{"type":45,"tag":76,"props":767,"children":768},{},[769,771,776],{"type":50,"value":770},"Bundle size grows with deployment size; on very large namespaces, scope with ",{"type":45,"tag":85,"props":772,"children":774},{"className":773},[],[775],{"type":50,"value":548},{"type":50,"value":777},".",{"type":45,"tag":76,"props":779,"children":780},{},[781,783,789],{"type":50,"value":782},"Does not validate disagg transport — use ",{"type":45,"tag":85,"props":784,"children":786},{"className":785},[],[787],{"type":50,"value":788},"dynamo-interconnect-check",{"type":50,"value":790}," for that.",{"type":45,"tag":53,"props":792,"children":793},{"id":39},[794],{"type":50,"value":795},"Troubleshooting",{"type":45,"tag":484,"props":797,"children":798},{},[799,820],{"type":45,"tag":488,"props":800,"children":801},{},[802],{"type":45,"tag":492,"props":803,"children":804},{},[805,810,815],{"type":45,"tag":496,"props":806,"children":807},{},[808],{"type":50,"value":809},"Symptom",{"type":45,"tag":496,"props":811,"children":812},{},[813],{"type":50,"value":814},"Likely cause",{"type":45,"tag":496,"props":816,"children":817},{},[818],{"type":50,"value":819},"Next step",{"type":45,"tag":511,"props":821,"children":822},{},[823,846,879,903],{"type":45,"tag":492,"props":824,"children":825},{},[826,836,841],{"type":45,"tag":518,"props":827,"children":828},{},[829,834],{"type":45,"tag":85,"props":830,"children":832},{"className":831},[],[833],{"type":50,"value":90},{"type":50,"value":835}," returns Forbidden on events\u002Fpods",{"type":45,"tag":518,"props":837,"children":838},{},[839],{"type":50,"value":840},"Service account lacks read RBAC",{"type":45,"tag":518,"props":842,"children":843},{},[844],{"type":50,"value":845},"Ask operator for read-only role binding on the namespace",{"type":45,"tag":492,"props":847,"children":848},{},[849,861,866],{"type":45,"tag":518,"props":850,"children":851},{},[852,854,859],{"type":50,"value":853},"Bundle missing ",{"type":45,"tag":85,"props":855,"children":857},{"className":856},[],[858],{"type":50,"value":103},{"type":50,"value":860}," status",{"type":45,"tag":518,"props":862,"children":863},{},[864],{"type":50,"value":865},"Operator not installed or different namespace",{"type":45,"tag":518,"props":867,"children":868},{},[869,871,877],{"type":50,"value":870},"Verify ",{"type":45,"tag":85,"props":872,"children":874},{"className":873},[],[875],{"type":50,"value":876},"dynamo-platform",{"type":50,"value":878}," operator is installed and watching the namespace",{"type":45,"tag":492,"props":880,"children":881},{},[882,893,898],{"type":45,"tag":518,"props":883,"children":884},{},[885,887],{"type":50,"value":886},"Model-download job in ",{"type":45,"tag":85,"props":888,"children":890},{"className":889},[],[891],{"type":50,"value":892},"Pending",{"type":45,"tag":518,"props":894,"children":895},{},[896],{"type":50,"value":897},"PVC unbound or HF secret missing",{"type":45,"tag":518,"props":899,"children":900},{},[901],{"type":50,"value":902},"Fix PVC binding or create the named HF secret, then rerun the job",{"type":45,"tag":492,"props":904,"children":905},{},[906,917,922],{"type":45,"tag":518,"props":907,"children":908},{},[909,911],{"type":50,"value":910},"Worker pods ",{"type":45,"tag":85,"props":912,"children":914},{"className":913},[],[915],{"type":50,"value":916},"CrashLoopBackOff",{"type":45,"tag":518,"props":918,"children":919},{},[920],{"type":50,"value":921},"Image\u002Fruntime mismatch or GPU not available",{"type":45,"tag":518,"props":923,"children":924},{},[925,927,933],{"type":50,"value":926},"Inspect container logs; check ",{"type":45,"tag":85,"props":928,"children":930},{"className":929},[],[931],{"type":50,"value":932},"nvidia.com\u002Fgpu",{"type":50,"value":934}," allocatable on nodes",{"type":45,"tag":53,"props":936,"children":938},{"id":937},"benchmark",[939],{"type":50,"value":940},"Benchmark",{"type":45,"tag":60,"props":942,"children":943},{},[944,946,952,954,960],{"type":50,"value":945},"See ",{"type":45,"tag":85,"props":947,"children":949},{"className":948},[],[950],{"type":50,"value":951},"BENCHMARK.md",{"type":50,"value":953}," for the NVCARPS-EVAL performance report (auto-generated by the NVSkills CI pipeline). To refresh, re-run ",{"type":45,"tag":85,"props":955,"children":957},{"className":956},[],[958],{"type":50,"value":959},"\u002Fnvskills-ci",{"type":50,"value":961}," on an upstream PR touching this skill.",{"type":45,"tag":53,"props":963,"children":965},{"id":964},"references",[966],{"type":50,"value":967},"References",{"type":45,"tag":72,"props":969,"children":970},{},[971,983],{"type":45,"tag":76,"props":972,"children":973},{},[974,976,981],{"type":50,"value":975},"Read ",{"type":45,"tag":85,"props":977,"children":979},{"className":978},[],[980],{"type":50,"value":290},{"type":50,"value":982}," for bucket-specific checks.",{"type":45,"tag":76,"props":984,"children":985},{},[986,987,992],{"type":50,"value":284},{"type":45,"tag":85,"props":988,"children":990},{"className":989},[],[991],{"type":50,"value":526},{"type":50,"value":993}," for read-only bundle collection.",{"type":45,"tag":995,"props":996,"children":997},"style",{},[998],{"type":50,"value":999},"html .light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html.light .shiki span {color: var(--shiki-light);background: var(--shiki-light-bg);font-style: var(--shiki-light-font-style);font-weight: var(--shiki-light-font-weight);text-decoration: var(--shiki-light-text-decoration);}html .default .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .shiki span {color: var(--shiki-default);background: var(--shiki-default-bg);font-style: var(--shiki-default-font-style);font-weight: var(--shiki-default-font-weight);text-decoration: var(--shiki-default-text-decoration);}html .dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}html.dark .shiki span {color: var(--shiki-dark);background: var(--shiki-dark-bg);font-style: var(--shiki-dark-font-style);font-weight: var(--shiki-dark-font-weight);text-decoration: var(--shiki-dark-text-decoration);}",{"items":1001,"total":1155},[1002,1020,1035,1046,1058,1070,1083,1097,1110,1121,1135,1144],{"slug":1003,"name":1003,"fn":1004,"description":1005,"org":1006,"tags":1007,"stars":1017,"repoUrl":1018,"updatedAt":1019},"nemoclaw-user-guide","retrieve NemoClaw documentation and configuration","Guides human users' AI agents to the NemoClaw docs MCP server and canonical Fern documentation in Markdown form. Use when users ask how to install, configure, operate, troubleshoot, secure, or learn NemoClaw with an AI coding assistant. Trigger keywords - nemoclaw docs, use nemoclaw with ai agent, nemoclaw mcp docs, nemoclaw install help, nemoclaw quickstart, nemoclaw markdown docs, llms.txt, agent skills.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1008,1011,1014],{"name":1009,"slug":1010,"type":15},"Documentation","documentation",{"name":1012,"slug":1013,"type":15},"MCP","mcp",{"name":1015,"slug":1016,"type":15},"Search","search",21777,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FNemoClaw","2026-07-20T06:00:01.461044",{"slug":1021,"name":1021,"fn":1022,"description":1023,"org":1024,"tags":1025,"stars":1032,"repoUrl":1033,"updatedAt":1034},"mcore-build-and-dependency","manage Megatron-LM development environments","Container-based dev environment setup and dependency management for Megatron-LM. Covers acquiring and launching the CI container, uv package management, and updating uv.lock.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1026,1029,1030],{"name":1027,"slug":1028,"type":15},"Containers","containers",{"name":13,"slug":14,"type":15},{"name":1031,"slug":573,"type":15},"Python",17049,"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002FMegatron-LM","2026-07-27T06:06:11.249662",{"slug":1036,"name":1036,"fn":1037,"description":1038,"org":1039,"tags":1040,"stars":1032,"repoUrl":1033,"updatedAt":1045},"mcore-bump-base-image","update NVIDIA PyTorch base images","Bump the NVIDIA PyTorch base image (`nvcr.io\u002Fnvidia\u002Fpytorch:YY.MM-py3`) used by Megatron-LM CI. Covers the two pin sites (GitHub CI in `docker\u002F.ngc_version.dev` and GitLab CI in `.gitlab\u002Fstages\u002F01.build.yml`), the post-bump CI loop (re-run functional tests, refresh golden values, mark broken tests), and the gotchas that bit PRs",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1041,1044],{"name":1042,"slug":1043,"type":15},"CI\u002FCD","ci-cd",{"name":13,"slug":14,"type":15},"2026-07-14T05:25:59.97109",{"slug":1047,"name":1047,"fn":1048,"description":1049,"org":1050,"tags":1051,"stars":1032,"repoUrl":1033,"updatedAt":1057},"mcore-cicd","manage CI\u002FCD pipelines for Megatron-LM","CI\u002FCD reference for Megatron-LM. Covers CI pipeline structure, PR scope labels, triggering internal GitLab CI (which force-pushes the current branch to a pull-request\u002FBRANCH ref — always dry-run and verify the destination first; never run against shared or protected branches), and CI failure investigation.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1052,1053,1054],{"name":1042,"slug":1043,"type":15},{"name":13,"slug":14,"type":15},{"name":1055,"slug":1056,"type":15},"GitHub","github","2026-07-27T06:06:12.278222",{"slug":1059,"name":1059,"fn":1060,"description":1061,"org":1062,"tags":1063,"stars":1032,"repoUrl":1033,"updatedAt":1069},"mcore-create-issue","investigate CI failures and create issues","Investigate a failing GitHub Actions run or job and create a GitHub issue for the failure.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1064,1065,1066],{"name":20,"slug":21,"type":15},{"name":1055,"slug":1056,"type":15},{"name":1067,"slug":1068,"type":15},"Triage","triage","2026-07-14T05:25:57.442089",{"slug":1071,"name":1071,"fn":1072,"description":1073,"org":1074,"tags":1075,"stars":1032,"repoUrl":1033,"updatedAt":1082},"mcore-linting-and-formatting","lint and format Megatron-LM code","Linting and formatting for Megatron-LM. Covers running autoformat.sh, tools (ruff, black, isort, pylint, mypy), and code style rules.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1076,1079],{"name":1077,"slug":1078,"type":15},"Best Practices","best-practices",{"name":1080,"slug":1081,"type":15},"Code Analysis","code-analysis","2026-07-14T05:25:56.18433",{"slug":1084,"name":1084,"fn":1085,"description":1086,"org":1087,"tags":1088,"stars":1032,"repoUrl":1033,"updatedAt":1096},"mcore-migrate-gpt-to-hybrid","migrate Megatron-LM models to HybridModel","Migration guide for moving Megatron Core GPTModel checkpoints, model providers, training commands, and layer mappings to HybridModel.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1089,1092,1095],{"name":1090,"slug":1091,"type":15},"Machine Learning","machine-learning",{"name":1093,"slug":1094,"type":15},"Migration","migration",{"name":9,"slug":8,"type":15},"2026-07-17T06:07:11.777011",{"slug":1098,"name":1098,"fn":1099,"description":1100,"org":1101,"tags":1102,"stars":1032,"repoUrl":1033,"updatedAt":1109},"mcore-onboard-gb200-1node-tests","onboard functional tests for GB200","Onboard 1-node GitHub MR functional tests for GB200 from existing mr-scoped 2-node tests.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1103,1106],{"name":1104,"slug":1105,"type":15},"QA","qa",{"name":1107,"slug":1108,"type":15},"Testing","testing","2026-07-14T05:25:53.673039",{"slug":1111,"name":1111,"fn":1112,"description":1113,"org":1114,"tags":1115,"stars":1032,"repoUrl":1033,"updatedAt":1120},"mcore-run-on-slurm","launch distributed training jobs on SLURM","How to launch distributed Megatron-LM training jobs on a SLURM cluster. Covers a minimal sbatch skeleton, environment-variable setup for torch.distributed.run, CUDA_DEVICE_MAX_CONNECTIONS rules across hardware and parallelism modes, container conventions, monitoring, and per-rank failure diagnosis.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1116,1117],{"name":13,"slug":14,"type":15},{"name":1118,"slug":1119,"type":15},"Infrastructure","infrastructure","2026-07-14T05:25:49.362534",{"slug":1122,"name":1122,"fn":1123,"description":1124,"org":1125,"tags":1126,"stars":1032,"repoUrl":1033,"updatedAt":1134},"mcore-split-pr","split pull requests to reduce review load","Split a PR into multiple PRs to reduce the number of required CODEOWNERS reviewer groups.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1127,1130,1131],{"name":1128,"slug":1129,"type":15},"Code Review","code-review",{"name":1055,"slug":1056,"type":15},{"name":1132,"slug":1133,"type":15},"Pull Requests","pull-requests","2026-07-14T05:26:01.226578",{"slug":1136,"name":1136,"fn":1137,"description":1138,"org":1139,"tags":1140,"stars":1032,"repoUrl":1033,"updatedAt":1143},"mcore-testing","run and manage Megatron-LM tests","Test system for Megatron-LM. Covers test layout, recipe YAML structure, adding and running unit and functional tests, golden values, marker filters, and CI parity.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1141,1142],{"name":1104,"slug":1105,"type":15},{"name":1107,"slug":1108,"type":15},"2026-07-14T05:25:54.928983",{"slug":1145,"name":1145,"fn":1146,"description":1147,"org":1148,"tags":1149,"stars":1032,"repoUrl":1033,"updatedAt":1154},"nightly-sync","manage nightly main-to-dev sync workflows","Domain knowledge for the nightly main-to-dev sync workflow. Covers merge strategy, CI architecture, failure investigation, and known issues.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1150,1153],{"name":1151,"slug":1152,"type":15},"Automation","automation",{"name":1042,"slug":1043,"type":15},"2026-07-30T05:29:03.275638",496,{"items":1157,"total":1253},[1158,1175,1185,1199,1209,1224,1239],{"slug":1159,"name":1159,"fn":1160,"description":1161,"org":1162,"tags":1163,"stars":22,"repoUrl":23,"updatedAt":1174},"accelerated-computing-cudf","accelerate data processing with cuDF","Official NVIDIA-authored guidance for NVIDIA cuDF GPU DataFrames, pandas acceleration, dask-cuDF, ETL, joins, groupby, CSV\u002FParquet I\u002FO, nullable semantics, and multi-GPU DataFrame workloads.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1164,1167,1170,1171],{"name":1165,"slug":1166,"type":15},"Data Analysis","data-analysis",{"name":1168,"slug":1169,"type":15},"Data Engineering","data-engineering",{"name":9,"slug":8,"type":15},{"name":1172,"slug":1173,"type":15},"Performance","performance","2026-07-14T05:28:43.176466",{"slug":1176,"name":1176,"fn":1177,"description":1178,"org":1179,"tags":1180,"stars":22,"repoUrl":23,"updatedAt":1184},"aiq-deploy","deploy and manage NVIDIA AI-Q infrastructure","Use when asked to install, deploy, run, validate, troubleshoot, or stop NVIDIA AI-Q Blueprint infrastructure.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1181,1182,1183],{"name":13,"slug":14,"type":15},{"name":1118,"slug":1119,"type":15},{"name":9,"slug":8,"type":15},"2026-07-14T05:29:06.667109",{"slug":1186,"name":1186,"fn":1187,"description":1188,"org":1189,"tags":1190,"stars":22,"repoUrl":23,"updatedAt":1198},"aiq-research","conduct deep research with AI-Q","Use when asked to run deep research or AI-Q research through a reachable NVIDIA AI-Q Blueprint backend.\n",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1191,1194,1195],{"name":1192,"slug":1193,"type":15},"Agents","agents",{"name":9,"slug":8,"type":15},{"name":1196,"slug":1197,"type":15},"Research","research","2026-07-14T05:28:06.816956",{"slug":1200,"name":1200,"fn":1201,"description":1202,"org":1203,"tags":1204,"stars":22,"repoUrl":23,"updatedAt":1208},"amc-run-sample-calibration","run AMC sample dataset calibration","Run end-to-end calibration on the shipped sample dataset (sdg_08_2_sample_data_010926.zip) against a running AMC microservice. Use when user says 'test sample dataset', 'run sample calibration', 'verify AMC install', or 'launch and test'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1205,1206,1207],{"name":1165,"slug":1166,"type":15},{"name":9,"slug":8,"type":15},{"name":1107,"slug":1108,"type":15},"2026-07-17T05:29:03.913266",{"slug":1210,"name":1210,"fn":1211,"description":1212,"org":1213,"tags":1214,"stars":22,"repoUrl":23,"updatedAt":1223},"amc-run-video-calibration","calibrate video datasets with AutoMagicCalib","Calibrate a new dataset from pre-recorded video files via the AutoMagicCalib REST API. Use when user has local MP4s and says 'calibrate my videos', 'run AMC on these videos', or similar. For RTSP\u002Flive streams, use amc-run-rtsp-calibration instead.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1215,1216,1219,1220],{"name":1151,"slug":1152,"type":15},{"name":1217,"slug":1218,"type":15},"Imaging","imaging",{"name":9,"slug":8,"type":15},{"name":1221,"slug":1222,"type":15},"Video","video","2026-07-17T05:28:53.905004",{"slug":1225,"name":1225,"fn":1226,"description":1227,"org":1228,"tags":1229,"stars":22,"repoUrl":23,"updatedAt":1238},"amc-setup-calibration-stack","deploy AutoMagicCalib microservice with Docker","Launch AutoMagicCalib microservice and web UI from NGC release images via Docker Compose. Use when user says 'deploy auto calibration', 'launch auto calibration', 'launch AMC', 'start MS+UI', or 'set up auto-magic-calib'. Requires NGC API key.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1230,1231,1234,1235],{"name":13,"slug":14,"type":15},{"name":1232,"slug":1233,"type":15},"Docker","docker",{"name":9,"slug":8,"type":15},{"name":1236,"slug":1237,"type":15},"Operations","operations","2026-07-17T05:28:56.913999",{"slug":1240,"name":1240,"fn":1241,"description":1242,"org":1243,"tags":1244,"stars":22,"repoUrl":23,"updatedAt":1252},"cudaq-guide","develop quantum applications with CUDA-Q","CUDA-Q onboarding guide for installation, test programs, GPU simulation, QPU hardware, and quantum applications.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":9},[1245,1246,1249],{"name":9,"slug":8,"type":15},{"name":1247,"slug":1248,"type":15},"Quantum Computing","quantum-computing",{"name":1250,"slug":1251,"type":15},"Simulation","simulation","2026-07-14T05:26:58.898253",305]