NVIDIA/topograph v0.3.0

v0.3.0

Repository: NVIDIA/topograph

Tag: v0.3.0

Published: 2026-03-24T18:38:28Z

Prerelease: no

Release notes:

What's Changed

• return accurate HTTP errors on topology requests by @dmitsh in https://github.com/NVIDIA/topograph/pull/177
• chore: improve code coverage by @dmitsh in https://github.com/NVIDIA/topograph/pull/178
• fix: add option to skip TLS verification when creating HTTP client by @dmitsh in https://github.com/NVIDIA/topograph/pull/181
• chore: add utility function to generate URL by @dmitsh in https://github.com/NVIDIA/topograph/pull/182
• fix: in the status observer, trigger topology discovery when pods are ready by @dmitsh in https://github.com/NVIDIA/topograph/pull/183
• feat: add option to select k8s nodes for topology config by @dmitsh in https://github.com/NVIDIA/topograph/pull/184
• feat: add support for NVLink domain discovery in NetQ provider by @dmitsh in https://github.com/NVIDIA/topograph/pull/180
• feat: fetch OPIDs programmatically from NetQ API by @dmitsh in https://github.com/NVIDIA/topograph/pull/186
• Bump golang.org/x/crypto from 0.42.0 to 0.45.0 by @dependabot[bot] in https://github.com/NVIDIA/topograph/pull/185
• chore: update Nebius docs by @dmitsh in https://github.com/NVIDIA/topograph/pull/187
• fix: in Slinky, locate the login pod in the Running state by @dmitsh in https://github.com/NVIDIA/topograph/pull/188
• chore: return an error if the DRA provider has an empty domain set by @dmitsh in https://github.com/NVIDIA/topograph/pull/189
• chore: add docker image for ib tools by @dmitsh in https://github.com/NVIDIA/topograph/pull/190
• chore(k8s): update values file for deployment with InfiniBand provider by @dmitsh in https://github.com/NVIDIA/topograph/pull/191
• fix(topograph): the 'topology' endpoint to return '102 Processing' when the request has't completed yet by @dmitsh in https://github.com/NVIDIA/topograph/pull/192
• fix(topograph): golang does not support '102 Processing', fallback to '202 Accepted' by @dmitsh in https://github.com/NVIDIA/topograph/pull/193
• fix(topograph): update README by @dmitsh in https://github.com/NVIDIA/topograph/pull/194
• Update k8s.md image reference by @resker in https://github.com/NVIDIA/topograph/pull/195
• fix(topograph) fix retry logic by @dmitsh in https://github.com/NVIDIA/topograph/pull/196
• chore(topograph) simplify http helper functions by @dmitsh in https://github.com/NVIDIA/topograph/pull/197
• feat(lambdai) implement initial draft of topology provider by @dmitsh in https://github.com/NVIDIA/topograph/pull/198
• chore(k8s) add ServiceMonitor by @dmitsh in https://github.com/NVIDIA/topograph/pull/199
• feat(lambdai): implement provider simulator by @ravisoundar in https://github.com/NVIDIA/topograph/pull/200
• chore(k8s) upgrade to Kubernetes 1.34 by @dmitsh in https://github.com/NVIDIA/topograph/pull/201
• feat(slurm) add support for dynamic nodes by @dmitsh in https://github.com/NVIDIA/topograph/pull/202
• feat(gcp): allow external API authentication by @dmitsh in https://github.com/NVIDIA/topograph/pull/204
• feat(topograph) add payloads for integration tests by @dmitsh in https://github.com/NVIDIA/topograph/pull/203
• fix(slurm): fix block order by @dmitsh in https://github.com/NVIDIA/topograph/pull/207
• chore(topograph): rename backend switch tiers by @dmitsh in https://github.com/NVIDIA/topograph/pull/208
• fix(slurm): revert dynamic node usage by @dmitsh in https://github.com/NVIDIA/topograph/pull/206
• feat(topograph): Integration Test Support by @ravisoundar in https://github.com/NVIDIA/topograph/pull/205
• fix(topograph): log an error for HTTP error status codes by @dmitsh in https://github.com/NVIDIA/topograph/pull/209
• fix(topograph): return HTTP 202 for requests in progress by @dmitsh in https://github.com/NVIDIA/topograph/pull/210
• chore(topograph): include topograph version in metrics by @dmitsh in https://github.com/NVIDIA/topograph/pull/211
• feat: replace model_path with modelFileName by @dmitsh in https://github.com/NVIDIA/topograph/pull/212
• chore(topograph): update otel to v1.40.0 by @dmitsh in https://github.com/NVIDIA/topograph/pull/213
• chore(aws): upgrade SDK version to support Secondary Networks by @dmitsh in https://github.com/NVIDIA/topograph/pull/214
• fix(topograph): tune topology request latency buckets for better percentile accuracy by @dmitsh in https://github.com/NVIDIA/topograph/pull/215
• feat: Request Aggregation based on payload by @ravisoundar in https://github.com/NVIDIA/topograph/pull/217
• feat: lookup endpoint by @ravisoundar in https://github.com/NVIDIA/topograph/pull/218
• chore(topograph): use FNV-64 for payload hashing by @dmitsh in https://github.com/NVIDIA/topograph/pull/219
• chore(k8s): set topology node label names only if defined by @dmitsh in https://github.com/NVIDIA/topograph/pull/221
• fix(gcp): allow getting project ID from credentials by @dmitsh in https://github.com/NVIDIA/topograph/pull/220
• chore: Updated helm releases to keep existing files by @ravisoundar in https://github.com/NVIDIA/topograph/pull/222
• fix(topograph): allow credentials to contain 'any' values by @dmitsh in https://github.com/NVIDIA/topograph/pull/223
• feat(gcp): enable k8s deployment with federated workload identity by @dmitsh in https://github.com/NVIDIA/topograph/pull/224
• chore(helm): update version to 0.3.0 by @dmitsh in https://github.com/NVIDIA/topograph/pull/226
• chore(gcp): Documentation for GCP workload identity federation by @ravisoundar in https://github.com/NVIDIA/topograph/pull/225
• Nebius GoSDK integration enhancement by @complynx in https://github.com/NVIDIA/topograph/pull/227
• chore(topograph): clean up go.[mod|sum] by @dmitsh in https://github.com/NVIDIA/topograph/pull/228
• fix(topograph): remove node from partition topology when node data is missing by @dmitsh in https://github.com/NVIDIA/topograph/pull/229
• fix(k8s): update resource requests and limits by @dmitsh in https://github.com/NVIDIA/topograph/pull/231
• fix(topograph): fix typo in Slurm YAML format by @dmitsh in https://github.com/NVIDIA/topograph/pull/232
• feat: Add default provider and engine params by @ravisoundar in https://github.com/NVIDIA/topograph/pull/234
• feat(topograph): support trimming upper network tiers by @dmitsh in https://github.com/NVIDIA/topograph/pull/233
• feat(slurm): default to flat YAML topology if partition topology is absent by @dmitsh in https://github.com/NVIDIA/topograph/pull/235
• chore(deps): Bump google.golang.org/grpc from…