v1.6.0-rc.1

NVIDIA Fleet Intelligence Agent v1.6.0-rc.1

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

New Features

• 974c1e6: feat: add DCGM-backed GPU inventory provider (#251) (@ooolafhu)
• a96df29: feat: add opt-in GPU workload attribution (#258) (@ooolafhu)
• ebad8f0: feat: support GPUCluster DCGM discovery (#261) (@ooolafhu)
• ca9dfcf: feat: support fault injection configuration through Helm (#273) (@Brilly-Bohyun)

Bug Fixes

• a45d97f: fix(deps): update gRPC for GO-2026-6348 (#260) (@ooolafhu)
• e954a6a: fix(packaging): install DCGM HostEngine dependency (#259) (@ooolafhu)
• 67720dc: fix: make exported health data deterministically ordered (#257) (@rsampaio)
• fb92533: fix: prevent invalid Slurm job IDs from panicking (#270) (@ooolafhu)

Documentation Updates

• 6eace17: docs: add OSS health score badge (#278) (@rsampaio)
• 081f838: docs: add publish-ready project guidance (#265) (@rsampaio)
• 05e024b: docs: document issue prioritization (#275) (@rsampaio)
• 14890a3: docs: document monitoring component contributions (#274) (@rsampaio)
• 1f0728a: docs: document release process (#269) (@rsampaio)
• b270f22: docs: explain agent runtime architecture (#272) (@rsampaio)
• 4d1383e: docs: link code of conduct from contributing guide (#266) (@rsampaio)
• 79c274f: docs: link public community discussions (#267) (@rsampaio)
• b5f165f: docs: require CI on release commit (#271) (@rsampaio)

Refactors

• bc9df44: refactor: migrate GPU inventory workflows from NVML (#253) (@ooolafhu)
• 2db2be2: refactor: migrate health components off NVML (#252) (@ooolafhu)
• 1ba38d0: refactor: remove direct NVML dependency (#254) (@ooolafhu)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/dsx-ai-factory/fleet-intelligence-agent/issues

v1.5.1-rc.1

NVIDIA Fleet Intelligence Agent v1.5.1-rc.1

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

New Features

• 974c1e6: feat: add DCGM-backed GPU inventory provider (#251) (@ooolafhu)

Bug Fixes

• a45d97f: fix(deps): update gRPC for GO-2026-6348 (#260) (@ooolafhu)
• e954a6a: fix(packaging): install DCGM HostEngine dependency (#259) (@ooolafhu)
• 67720dc: fix: make exported health data deterministically ordered (#257) (@rsampaio)

Refactors

• bc9df44: refactor: migrate GPU inventory workflows from NVML (#253) (@ooolafhu)
• 2db2be2: refactor: migrate health components off NVML (#252) (@ooolafhu)
• 1ba38d0: refactor: remove direct NVML dependency (#254) (@ooolafhu)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/dsx-ai-factory/fleet-intelligence-agent/issues

v1.5.0

NVIDIA Fleet Intelligence Agent v1.5.0

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/dsx-ai-factory/fleet-intelligence-agent/issues

What's Changed

• fix: bump the go crypto version to fix cve by @ambermingxin in #241
• chore: add linux package and checksum signing by @jingxiang-z in #236
• fix: Add trigger for docs in release. by @ebalduf in #243
• ci: pin goreleaser-action to v5.1.0 SHA by @kshivakumar-nvidia in #244
• chore: update repository organization references by @jingxiang-z in #245
• fix(helm): mount host identity files during enrollment by @jingxiang-z in #247
• feat: sync resolved node membership by @jingxiang-z in #246
• chore: rename agent Go module by @jingxiang-z in #248
• fix: set default log level to warn by @jingxiang-z in #250
• docs: add enrollment endpoints by @ebalduf in #256

New Contributors

• @kshivakumar-nvidia made their first contribution in #244

Full Changelog: v1.4.0...v1.5.0

v1.5.0-rc.1

NVIDIA Fleet Intelligence Agent v1.5.0-rc.1

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

New Features

• b580112: feat: sync resolved node membership (#246) (@jingxiang-z)

Bug Fixes

• 8ebce32: fix(helm): mount host identity files during enrollment (#247) (@jingxiang-z)
• 7240661: fix: Add trigger for docs in release. (#243) (@ebalduf)
• 416ebf7: fix: bump the go crypto version to fix cve (#241) (@ambermingxin)
• 8dd8826: fix: set default log level to warn (#250) (@jingxiang-z)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/dsx-ai-factory/fleet-intelligence-agent/issues

v1.4.1-test.4

NVIDIA Fleet Intelligence Agent v1.4.1-test.4

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

Bug Fixes

• f9d10dd: fix(helm): mount host identity files during enrollment (@jingxiang-z)
• 7240661: fix: Add trigger for docs in release. (#243) (@ebalduf)
• 416ebf7: fix: bump the go crypto version to fix cve (#241) (@ambermingxin)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/dsx-ai-factory/fleet-intelligence-agent/issues

v1.4.1-test.3

NVIDIA Fleet Intelligence Agent v1.4.1-test.3

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

Bug Fixes

• 7240661: fix: Add trigger for docs in release. (#243) (@ebalduf)
• 416ebf7: fix: bump the go crypto version to fix cve (#241) (@ambermingxin)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/dsx-ai-factory/fleet-intelligence-agent/issues

v1.4.1-test.2

NVIDIA Fleet Intelligence Agent v1.4.1-test.2

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/NVIDIA/fleet-intelligence-agent/issues

v1.4.1-test.1

NVIDIA Fleet Intelligence Agent v1.4.1-test.1

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

New Features

• 06332a3: feat(release): sign Linux packages with 3S (@jingxiang-z)
• 54dbff5: feat(release): sign checksum manifest (@jingxiang-z)

Bug Fixes

• 5a3195a: fix(ci): install nvsec with pip resolver (@jingxiang-z)
• 890ee5b: fix(ci): install portable Python for nvsec (@jingxiang-z)
• ce4e5ce: fix(ci): locate draft releases reliably (@jingxiang-z)
• fb4cfe3: fix(release): address signature review findings (@jingxiang-z)
• fd114d4: fix(release): preserve unsigned artifact checksums (@jingxiang-z)
• 416ebf7: fix: bump the go crypto version to fix cve (#241) (@ambermingxin)

Documentation Updates

• 8a0f057: docs: make package verification optional (@jingxiang-z)
• abd471d: docs: remove maintainer signing details (@jingxiang-z)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Confirm that fleet-intelligence.pub.asc has fingerprint
FE0C 8B74 CA66 357C 13BE 197D CCE3 C963 0871 99B3, then authenticate
the checksum manifest:

gpg --show-keys --fingerprint fleet-intelligence.pub.asc
gpg --import fleet-intelligence.pub.asc
gpg --verify checksums.txt.asc checksums.txt

Support

For issues and questions, please visit: https://github.com/NVIDIA/fleet-intelligence-agent/issues

v1.4.0

NVIDIA Fleet Intelligence Agent v1.4.0

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Support

For issues and questions, please visit: https://github.com/NVIDIA/fleet-intelligence-agent/issues

What's Changed

• fix: preserve counter metric types in OTLP by @jingxiang-z in #198
• docs: update supported platform docs by @jingxiang-z in #207
• feat: add validation for ng/cz input by @ambermingxin in #206
• chore: disable accelerator-nvidia-processes by default by @ambermingxin in #209
• ci: harden github actions workflows by @jingxiang-z in #205
• fix: allow user to only set one field for ng or cz by @ambermingxin in #211
• fix: remove warning message by @ambermingxin in #214
• chore: update go version to 1.26.4 by @jingxiang-z in #215
• fix: OTLP export silently skips when endpoints are configured but no auth token is present. by @miconeilaws in #199
• fix: ubuntu deb package upgrade restart the service by @jingxiang-z in #218
• fix: apt remove preserve service enable by @jingxiang-z in #221
• fix: make DCGM group name configurable and different for each process by @jingxiang-z in #220
• fix: degrade dcgm query timeouts by @jingxiang-z in #222
• fix: include metric scrape interval in agent config by @jingxiang-z in #223
• feat: mark persistence mode diable as warning level by @ambermingxin in #225
• fix: initialize node uuid during enrollment by @jingxiang-z in #224
• feat: Inband attestation should collect stderr from nvattest sdk by @ambermingxin in #228
• chore: Update OpenTelemetry dependencies to v1.44.0 by @rsampaio in #230
• chore: add readme for agent local db to be persistent by @ambermingxin in #232
• fix(inventory): make change hash order invariant by @ooolafhu in #233
• fix: export the component state error and suggested actions to backend by @ambermingxin in #234
• fix: bump grpc to v1.82.1 by @rsampaio in #237
• docs: add Fern scaffolding and publish agent docs by @ebalduf in #238
• fix: fern docs url by @ebalduf in #239
• feat: add authenticated OpenTelemetry gateway by @rsampaio in #231
• feat: enrich OTLP telemetry identity by @jingxiang-z in #240
• fix: bump the go crypto version to fix cve (#241) by @ambermingxin in #242

New Contributors

• @miconeilaws made their first contribution in #199
• @ooolafhu made their first contribution in #233

Full Changelog: v1.3.1...v1.4.0

v1.4.0-rc2

NVIDIA Fleet Intelligence Agent v1.4.0-rc2

This release includes a host agent for GPU telemetry collection and attestation.

Available Packages

• DEB packages: amd64 and arm64 for Debian/Ubuntu systems
• RPM packages: x86_64 and aarch64 for RHEL/CentOS/Fedora systems
• Binary archives: Standalone binaries for both architectures

Installation

Download the appropriate package for your system from the assets below. For installation instructions, see the project README.

Changelog

Bug Fixes

• 3832ce2: fix: bump the go crypto version to fix cve (@ambermingxin)

Verification

All release artifacts are checksummed. Verify downloads using:

sha256sum -c checksums.txt

Support

For issues and questions, please visit: https://github.com/NVIDIA/fleet-intelligence-agent/issues