You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
nvdp deployed via helm chart v0.12.2 with gfd enabled, no other changes to values.yaml. Running on RHEL 8 with selinux enabled.
No nvidia.com/xxx labels are added to the kubernetes worker node. I have a workaround described in "2. Steps to reproduce the issue. " Please advice how to do it the right way.
nfd-worker logs permission denied (no sealert messages in system logs) on /host-sys/fs/selinux/enforce and /etc/kubernetes/node-feature-discovery/features.d//gfd
Result is that only nfd labels are added to the node, but no gfd labels.
I0729 12:08:26.342300 1 nfd-worker.go:155] Node Feature Discovery Worker v0.11.0
I0729 12:08:26.342435 1 nfd-worker.go:156] NodeName: 'gpu003.lab.cortical.io'
I0729 12:08:26.343127 1 nfd-worker.go:423] configuration file "/etc/kubernetes/node-feature-discovery/nfd-worker.conf" parsed
I0729 12:08:26.343265 1 nfd-worker.go:461] worker (re-)configuration successfully completed
I0729 12:08:26.343326 1 base.go:127] connecting to nfd-master at nvdp-node-feature-discovery-master:8080 ...
I0729 12:08:26.343376 1 component.go:36] [core]parsed scheme: ""
I0729 12:08:26.343389 1 component.go:36] [core]scheme "" not registered, fallback to default scheme
I0729 12:08:26.343416 1 component.go:36] [core]ccResolverWrapper: sending update to cc: {[{nvdp-node-feature-discovery-master:8080 <nil> 0 <nil>}] <nil> <nil>}
I0729 12:08:26.343430 1 component.go:36] [core]ClientConn switching balancer to "pick_first"
I0729 12:08:26.343439 1 component.go:36] [core]Channel switches to new LB policy "pick_first"
I0729 12:08:26.343494 1 component.go:36] [core]Subchannel Connectivity change to CONNECTING
I0729 12:08:26.343538 1 component.go:36] [core]Subchannel picks a new address "nvdp-node-feature-discovery-master:8080" to connect
I0729 12:08:26.343994 1 component.go:36] [core]Channel Connectivity change to CONNECTING
I0729 12:08:26.346260 1 component.go:36] [core]Subchannel Connectivity change to READY
I0729 12:08:26.346296 1 component.go:36] [core]Channel Connectivity change to READY
W0729 12:08:26.361713 1 kernel.go:145] failed to detect the status of selinux: open /host-sys/fs/selinux/enforce: permission denied
E0729 12:08:26.361921 1 local.go:87] unable to access /etc/kubernetes/node-feature-discovery/features.d/: lstat /etc/kubernetes/node-feature-discovery/features.d//gfd: permission denied
I0729 12:08:26.436158 1 nfd-worker.go:472] starting feature discovery...
I0729 12:08:26.436813 1 nfd-worker.go:484] feature discovery completed
I0729 12:08:26.436833 1 nfd-worker.go:565] sending labeling request to nfd-master
2. Steps to reproduce the issue
Deploy via helm and check nfd-worker pod logs.
I have checked node-feature-discovery-worker Daemon Set definition and worker container has this security context defined:
This issue is stale because it has been open 90 days with no activity. This issue will be closed in 30 days unless new comments are made or the stale label is removed.
1. Issue or feature description
nvdp deployed via helm chart v0.12.2 with gfd enabled, no other changes to values.yaml. Running on RHEL 8 with selinux enabled.
No nvidia.com/xxx labels are added to the kubernetes worker node. I have a workaround described in "2. Steps to reproduce the issue. " Please advice how to do it the right way.
nfd-worker logs permission denied (no sealert messages in system logs) on
/host-sys/fs/selinux/enforce
and/etc/kubernetes/node-feature-discovery/features.d//gfd
Result is that only nfd labels are added to the node, but no gfd labels.
2. Steps to reproduce the issue
Deploy via helm and check nfd-worker pod logs.
I have checked node-feature-discovery-worker Daemon Set definition and worker container has this security context defined:
When I edit it and add
privileged: true
(and removeallowPrivilegeEscalation: false
), then it works and adds node labels nvidia.com/xxx:Additional information that might help better understand your environment and reproduce the bug:
RHEL 8.5
selinux enabled (container-selinux installed)
kubernetes 1.24.2
cri-o 1.24.1
Added selinux policy modules to enable nfd-worker and nvidia-device-plugin running without generating sealert logs:
allow container_t kubernetes_file_t:dir read;
allow container_t container_runtime_t:unix_stream_socket connectto;
allow container_t container_runtime_tmpfs_t:file { open read };
allow container_t xserver_misc_device_t:chr_file { getattr ioctl open read write };
The text was updated successfully, but these errors were encountered: