From 3e5408c4540cbff3be1231c84080ff4c1936111a Mon Sep 17 00:00:00 2001 From: Ali Sattari Date: Thu, 9 Jul 2026 16:28:00 +0200 Subject: [PATCH 1/2] update NCCL to 2.30.7-1 --- helm/soperator-activechecks/values.yaml | 2 +- images/jail/jail.dockerfile | 3 +-- images/sansible/sansible.dockerfile | 2 +- 3 files changed, 3 insertions(+), 4 deletions(-) diff --git a/helm/soperator-activechecks/values.yaml b/helm/soperator-activechecks/values.yaml index 98807eae9..c3096ffff 100644 --- a/helm/soperator-activechecks/values.yaml +++ b/helm/soperator-activechecks/values.yaml @@ -47,7 +47,7 @@ images: activeCheckImageTags: # Image training_diag from PR https://github.com/nebius/ml-containers/pull/88 "12.9.0": "12.9.0-ubuntu24.04-20260605082528" - "13.0.2": "13.0.2-ubuntu24.04-20260605082528" + "13.0.2": "13.0.2-ubuntu24.04-20260709140028" # Optional override in Pyxis ("reg#repo:tag") format. activeCheckImage: "" slurmJob: diff --git a/images/jail/jail.dockerfile b/images/jail/jail.dockerfile index 8482dbc19..4fab96537 100644 --- a/images/jail/jail.dockerfile +++ b/images/jail/jail.dockerfile @@ -3,7 +3,7 @@ ARG CUDA_VERSION ARG SLURM_VERSION # https://github.com/nebius/ml-containers/pull/88 -FROM cr.eu-north1.nebius.cloud/ml-containers/slurm_training_diag:slurm${SLURM_VERSION}-cuda${CUDA_VERSION}-ubuntu24.04-20260605082538 AS jail +FROM cr.eu-north1.nebius.cloud/ml-containers/slurm_training_diag:slurm${SLURM_VERSION}-cuda${CUDA_VERSION}-ubuntu24.04-20260709135942 AS jail # Create directory for pivoting host's root RUN mkdir -m 555 /mnt/host @@ -128,4 +128,3 @@ COPY --from=untaped /jail_restic /jail_restic COPY images/jail/populate_jail_entrypoint.sh /opt/bin/ RUN chmod +x /opt/bin/populate_jail_entrypoint.sh ENTRYPOINT ["/opt/bin/populate_jail_entrypoint.sh"] - diff --git a/images/sansible/sansible.dockerfile b/images/sansible/sansible.dockerfile index e310ff530..6b5656b7b 100644 --- a/images/sansible/sansible.dockerfile +++ b/images/sansible/sansible.dockerfile @@ -1,7 +1,7 @@ # syntax=docker.io/docker/dockerfile-upstream:1.20.0 # https://github.com/nebius/ml-containers/pull/88 -FROM cr.eu-north1.nebius.cloud/ml-containers/ansible_roles:noble-20260605082515 AS sansible +FROM cr.eu-north1.nebius.cloud/ml-containers/ansible_roles:noble-20260709135925 AS sansible # Install common packages RUN apt update && \ From 311e176827086df58a8c4aeddb3a98bc859d5139 Mon Sep 17 00:00:00 2001 From: Ali Sattari Date: Fri, 10 Jul 2026 09:55:01 +0200 Subject: [PATCH 2/2] make Pasha happy --- helm/soperator-activechecks/values.yaml | 1 + images/jail/jail.dockerfile | 2 +- images/sansible/sansible.dockerfile | 2 +- 3 files changed, 3 insertions(+), 2 deletions(-) diff --git a/helm/soperator-activechecks/values.yaml b/helm/soperator-activechecks/values.yaml index c3096ffff..773071143 100644 --- a/helm/soperator-activechecks/values.yaml +++ b/helm/soperator-activechecks/values.yaml @@ -47,6 +47,7 @@ images: activeCheckImageTags: # Image training_diag from PR https://github.com/nebius/ml-containers/pull/88 "12.9.0": "12.9.0-ubuntu24.04-20260605082528" + # Image training_diag from PR https://github.com/nebius/ml-containers/pull/92 "13.0.2": "13.0.2-ubuntu24.04-20260709140028" # Optional override in Pyxis ("reg#repo:tag") format. activeCheckImage: "" diff --git a/images/jail/jail.dockerfile b/images/jail/jail.dockerfile index 4fab96537..28765d0fb 100644 --- a/images/jail/jail.dockerfile +++ b/images/jail/jail.dockerfile @@ -2,7 +2,7 @@ ARG CUDA_VERSION ARG SLURM_VERSION -# https://github.com/nebius/ml-containers/pull/88 +# https://github.com/nebius/ml-containers/pull/92 FROM cr.eu-north1.nebius.cloud/ml-containers/slurm_training_diag:slurm${SLURM_VERSION}-cuda${CUDA_VERSION}-ubuntu24.04-20260709135942 AS jail # Create directory for pivoting host's root diff --git a/images/sansible/sansible.dockerfile b/images/sansible/sansible.dockerfile index 6b5656b7b..5b563ca0b 100644 --- a/images/sansible/sansible.dockerfile +++ b/images/sansible/sansible.dockerfile @@ -1,6 +1,6 @@ # syntax=docker.io/docker/dockerfile-upstream:1.20.0 -# https://github.com/nebius/ml-containers/pull/88 +# https://github.com/nebius/ml-containers/pull/92 FROM cr.eu-north1.nebius.cloud/ml-containers/ansible_roles:noble-20260709135925 AS sansible # Install common packages