Name and Version
When I upgraded my k8s Deployment to image with tag server-cuda-b9282, The server didn't start and logged the following error:
/app/llama-server: error while loading shared libraries: libllama-common.so.0: cannot open shared object file: No such file or directory
I reverted back to previous image with tag server-cuda-b9265 which still works.
Operating systems
Linux
Which llama.cpp modules do you know to be affected?
llama-server
Command line
- args:
- --host
- 0.0.0.0
- --metrics
- --models-max
- "1"
- --models-preset
- /etc/llama/config.ini
- --n-gpu-layers
- "999"
- --numa
- isolate
- --parallel
- "1"
- --port
- "8080"
- --threads
- "16"
- --threads-batch
- "16"Problem description & steps to reproduce
Run the new image.
First Bad Commit
No response
Relevant log output
Logs
_msg /app/llama-server: error while loading shared libraries: libllama-common.so.0: cannot open shared object file: No such file or directory
_stream {kubernetes.container_name="llama-cpp",kubernetes.pod_name="llama-llm-6757954f8b-ll4f2",kubernetes.pod_namespace="ai",stream="stderr"}
_stream_id 000000000000000037c30d12a913461f3b0a0553eb7a1d43
_time 2026-05-22T08:59:34.388529214Z
file /var/log/pods/ai_llama-llm-6757954f8b-ll4f2_1b22b0f9-b128-4624-8d96-b4059de6afbd/llama-cpp/1.log
kubernetes.container_id cri-o://db851fe31fa1d4f01569d1062359fd39179cfbd1a2c2255f9167eab2556fbcb5
kubernetes.container_image ghcr.io/ggml-org/llama.cpp:server-cuda
kubernetes.container_image_id ghcr.io/ggml-org/llama.cpp@sha256:226145e87d0064f34cd365432e74da5213a47f7203f3ba45e72fadc199b16194
kubernetes.container_name llama-cpp
kubernetes.namespace_labels.kubernetes.io/metadata.name ai
kubernetes.namespace_labels.name ai
kubernetes.node_labels.kubernetes.io/arch amd64
kubernetes.node_labels.kubernetes.io/hostname nk8s4.marand.si
kubernetes.node_labels.kubernetes.io/os linux
kubernetes.pod_annotations.cni.projectcalico.org/containerID 4ea5791697debddfa032a395b09cca5e91dfb852bd9fa18de5e585826f57ecdb
kubernetes.pod_annotations.cni.projectcalico.org/podIP 10.160.212.171/32
kubernetes.pod_annotations.cni.projectcalico.org/podIPs 10.160.212.171/32
kubernetes.pod_annotations.config-hash 71f197b38b152b7697d9b1ad33efc06ca5b7e341431e58d8ed0d8985eda1e40
kubernetes.pod_annotations.kubectl.kubernetes.io/restartedAt 2026-05-08T09:08:43.578Z
kubernetes.pod_ip 10.160.212.171
kubernetes.pod_ips ["10.160.212.171"]
kubernetes.pod_labels.app llama-llm
kubernetes.pod_labels.pod-template-hash 6757954f8b
kubernetes.pod_name llama-llm-6757954f8b-ll4f2
kubernetes.pod_namespace ai
kubernetes.pod_node_name nk8s4.marand.si
kubernetes.pod_owner ReplicaSet/llama-llm-6757954f8b
kubernetes.pod_uid 1b22b0f9-b128-4624-8d96-b4059de6afbd
level error
source_type kubernetes_logs
stream stderr
timestamp_end 2026-05-22T08:59:34.388529214Z
Name and Version
When I upgraded my k8s Deployment to image with tag server-cuda-b9282, The server didn't start and logged the following error:
/app/llama-server: error while loading shared libraries: libllama-common.so.0: cannot open shared object file: No such file or directory
I reverted back to previous image with tag server-cuda-b9265 which still works.
Operating systems
Linux
Which llama.cpp modules do you know to be affected?
llama-server
Command line
- args: - --host - 0.0.0.0 - --metrics - --models-max - "1" - --models-preset - /etc/llama/config.ini - --n-gpu-layers - "999" - --numa - isolate - --parallel - "1" - --port - "8080" - --threads - "16" - --threads-batch - "16"Problem description & steps to reproduce
Run the new image.
First Bad Commit
No response
Relevant log output
Logs