vrr/cozystack
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
cozystack/packages/system
History
Exact
Aleksei Sviridkin f45facca77
feat(hami): add HAMi GPU virtualization system package (#2484) 
## What this PR does

Documentation: https://github.com/cozystack/website/pull/517

Integrates [HAMi](https://github.com/Project-HAMi/HAMi) v2.8.1 (CNCF
Sandbox) into Cozystack as a system-level package for fractional GPU
sharing in tenant Kubernetes clusters.

The integration covers three layers:

- **System chart** (`packages/system/hami/`): Vendors upstream HAMi Helm
chart with device plugin, scheduler extender, mutating webhook, and
RuntimeClass configuration. The broken DRA subchart was removed — it
renders resources even when disabled and references unpublished images.

- **Kubernetes app addon** (`packages/apps/kubernetes/`): HAMi exposed
as an optional toggle (`hami.enabled`). Automatically disables GPU
Operator's native device plugin when active to avoid conflicts. Enforces
hard dependency on GPU Operator.

- **Platform registration** (`packages/core/platform/`): HAMi declared
as PackageSource with gpu-operator dependency, included in the iaas
bundle.

**Known limitation**: HAMi-core relies on a private glibc symbol
(`_dl_sym`) removed in glibc 2.34, which breaks compute isolation on
modern container images (Ubuntu 22.04+, PyTorch/TensorFlow official
images). Alpine/musl is entirely incompatible. See the package README
for details and upstream issue references.

### Screenshots

N/A — no UI changes.

### Release note

```release-note
feat(hami): add HAMi GPU virtualization as an optional system package for fractional GPU sharing. Enables memory and compute isolation for NVIDIA GPUs across tenant workloads. Integrates with GPU Operator and can be enabled per-cluster via the hami.enabled toggle. Note: compute isolation requires glibc < 2.34 in workload containers.
```

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Added HAMi GPU virtualization middleware as an optional cluster addon
to enable fractional GPU sharing.
* HAMi addon includes an enable toggle (default: disabled) and a
customizable Helm values override; when enabled it enforces GPU Operator
presence and adjusts GPU Operator values accordingly.
* New HelmRelease integration for deploying HAMi with dependency
ordering and conditional rendering.

* **Documentation**
* Added HAMi setup guide, compatibility notes, and updated cluster addon
configuration reference.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->
2026-04-29 16:22:58 +03:00
..
application-definition-crd refactor(api): rename CozystackResourceDefinition to ApplicationDefinition 2026-01-15 22:35:56 +01:00
backup-controller Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
backupstrategy-controller fix(backups): move velero-configmap Role to velero chart (#2459) 2026-04-23 11:57:54 +05:00
bootbox refactor(labels): remove cozystack.io/ui label 2026-01-19 13:59:29 +01:00
bootbox-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
bucket Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
bucket-rd fix(bucket): regenerate schema and docs from values.yaml annotations 2026-03-03 19:32:32 +03:00
capi-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
capi-providers-bootstrap refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
capi-providers-core refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
capi-providers-cpprovider fix(capi-provider): update Kamaji CAPI provider to v0.16.0 2026-02-19 22:12:33 +03:00
capi-providers-infraprovider refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
cert-manager [cert-manager] Added cleanup before crds upgrade 2026-02-26 18:32:35 +05:00
cert-manager-crds [cert-manager] Updated cert-manager to 1.19.3 with new crd packaging 2026-02-26 15:28:19 +03:00
cert-manager-issuers fix(platform): migrate ACME HTTP-01 to ingressClassName API 2026-04-21 17:00:31 +05:00
cilium fix(cilium): rebuild image multi-arch and pin tag to 1.19.3 2026-04-23 17:33:19 +03:00
cilium-networkpolicy refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
clickhouse-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
clickhouse-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
cluster-autoscaler feat(cluster-autoscaler): enable enforce-node-group-min-size by default 2026-02-14 11:42:52 +01:00
clustersecret-operator Add clustersecret-operator system package 2026-02-10 22:47:20 +05:00
coredns fix coredns serviceaccount to match kubernetes bootstrap rbac 2026-02-02 15:32:50 +01:00
cozy-proxy feat(cozy-proxy): bump to v0.3.0 2026-04-28 11:14:27 +02:00
cozystack-api Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
cozystack-basics [system] Fix tenant RBAC for endpointslices read access (#2257) 2026-03-27 17:48:46 +05:00
cozystack-controller Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
cozystack-scheduler feat(scheduler): storage-aware scheduling 2026-04-13 17:15:09 +03:00
dashboard Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
etcd-operator fix(etcd-operator): replace deprecated kube-rbac-proxy image 2026-03-10 08:30:43 +01:00
etcd-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
external-dns feat(extra): add external-dns as standalone extra package 2026-03-10 12:48:01 +01:00
external-dns-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
external-secrets-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
flux-plunger Fix pre-commit check 2026-01-20 03:24:32 +01:00
fluxcd refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
fluxcd-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
foundationdb-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
foundationdb-rd refactor(labels): remove cozystack.io/ui label 2026-01-19 13:59:29 +01:00
gateway-api-crds refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
goldpinger refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
gpu-operator docs(gpu-operator): clarify violation counter unit ambiguity in DCGM CSV 2026-04-28 12:07:31 +03:00
grafana-operator Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
hami chore(hami): document and partially automate vendor patches in Makefile 2026-04-29 11:14:41 +03:00
harbor Fix system postgresql images to 17.7-standard-trixie 2026-04-10 12:47:46 +05:00
harbor-rd fix(harbor): remove incorrect tenant module flags 2026-04-21 20:26:13 +02:00
hetzner-robotlb chore(hetzner-robotlb): update robotlb chart to appVersion 0.0.6 2026-04-23 16:57:23 +05:00
http-cache-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
info-rd refactor(labels): remove cozystack.io/ui label 2026-01-19 13:59:29 +01:00
ingress-nginx Replaced direct chart edits with patch files 2026-03-02 12:35:14 +05:00
ingress-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
kafka-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
kafka-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
kamaji fix(kamaji): increase memory limits and add startup probe (#2421) 2026-04-24 23:11:53 +03:00
keycloak fix(platform): migrate ACME HTTP-01 to ingressClassName API 2026-04-21 17:00:31 +05:00
keycloak-configure [keycloak-configure] Make values.yaml self-documenting 2026-04-02 19:29:54 +03:00
keycloak-operator [keycloak-operator] Update to v1.32.0 2026-03-11 15:15:09 +03:00
kilo [kilo] Switch from fork to upstream squat/kilo 2026-03-23 15:18:23 +03:00
kubeovn fix(kube-ovn): bump kube-ovn to v1.15.10 with port-group regression fix 2026-04-21 20:09:19 +02:00
kubeovn-plunger Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
kubeovn-webhook Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
kubernetes-rd chore(kubernetes): regenerate CRD after merge conflict resolution 2026-04-28 19:02:35 +03:00
kubevirt fix(kubevirt): disable serial console log to fix VM pod initialization 2026-02-21 20:46:31 +01:00
kubevirt-cdi feat(cdi): switch clone strategy to csi-clone 2026-03-10 12:42:01 +01:00
kubevirt-cdi-operator [cdi-operator] Update to v1.64.0 2026-02-20 20:16:33 +01:00
kubevirt-csi-node Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
kubevirt-instancetypes refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
kubevirt-operator [kubevirt-operator] Update to v1.6.4 2026-02-20 20:16:34 +01:00
lineage-controller-webhook refactor(lineage-controller-webhook): align with cozystack-api shape 2026-04-28 14:55:11 +03:00
linstor feat(linstor): bump linstor-csi to v1.10.6 with Protocol-C dual-attach fix 2026-04-27 14:50:47 +02:00
linstor-gui build(linstor): use shared BUILDX_ARGS for linstor-gui image build 2026-04-27 17:31:07 +05:00
linstor-scheduler feat(scheduler): storage-aware scheduling 2026-04-13 17:15:09 +03:00
local-ccm Update local-ccm v0.3.0 2026-02-10 20:19:03 +01:00
mariadb-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
mariadb-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
metallb Prepare release v1.0.0-alpha.1 2026-01-15 23:32:41 +00:00
metrics-server refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
mongodb-operator Fix pre-commit check 2026-01-20 03:24:32 +01:00
mongodb-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
monitoring feat(monitoring): add GPU observability dashboards and recording rules (#2418) 2026-04-28 22:42:01 +03:00
monitoring-agents fix(monitoring): prevent many-to-many match in util-per-watt recording rule 2026-04-28 11:21:12 +03:00
monitoring-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
multus Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
nats Add monitoring for NATs 2026-02-17 22:54:12 +01:00
nats-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
nfs-driver refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
objectstorage-controller Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
openbao feat(openbao): add system chart with vendored upstream 2026-02-17 20:03:01 +03:00
openbao-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
opencost Fix pre-commit check 2026-01-20 03:24:32 +01:00
opensearch-operator [opensearch] Address PR review feedback 2026-03-17 11:20:44 +01:00
opensearch-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
piraeus-operator [piraeus-operator] Split LINSTOR controller availability and metrics alerts 2026-03-25 16:11:14 +03:00
piraeus-operator-crds Move CRDs installation logic to piraeus-operator-crds chart 2026-02-11 22:54:56 +01:00
postgres-operator docs(postgres-operator): fix backoffLimit/activeDeadlineSeconds comment math 2026-04-24 13:33:07 +03:00
postgres-rd docs(postgres): clarify bootstrap field descriptions 2026-04-21 14:34:30 +03:00
prometheus-operator-crds refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
qdrant [qdrant] Add Qdrant vector database application 2026-02-06 13:30:58 +03:00
qdrant-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
rabbitmq-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
rabbitmq-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
redis-operator refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
redis-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
reloader refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
seaweedfs Prepare release v1.3.0 2026-04-22 07:28:09 +00:00
seaweedfs-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
snapshot-controller refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
tcp-balancer-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
telepresence refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
tenant-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
velero fix(backups): move velero-configmap Role to velero chart 2026-04-23 10:41:41 +05:00
vertical-pod-autoscaler Post upgrade fixes for v1.0 packagesources 2026-01-28 20:08:35 +03:00
vertical-pod-autoscaler-crds refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
victoria-metrics-operator feat(monitoring): upgrade victoria-metrics-operator to v0.68.4 2026-04-20 15:48:53 +03:00
virtualprivatecloud-rd [docs] Update cozyvalues-gen 2026-03-25 15:59:22 +05:00
vm-default-images [vm-default-images] Default to replicated storageClass and document storage requirements 2026-04-14 19:59:20 +05:00
vm-disk-rd [vm-disk] Fix vm-disk-rd 2026-04-14 19:59:19 +05:00
vm-instance-rd [vm-instance] Add externalAllowICMP knob, drop in-PR changelog 2026-04-28 08:37:13 +02:00
vpn-rd [docs] Updated app go types 2026-03-25 15:57:25 +05:00
vsnap-crd refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00
Makefile refactor: move scripts to hack directory 2026-01-15 16:06:56 +01:00