# solo — shell
docker run -d --entrypoint '' --gpus all --ipc=host --shm-size=32gb --network=host --user 1000:1000 -v /etc/passwd:/etc/passwd:ro -v /etc/group:/etc/group:ro --security-opt no-new-privileges --security-opt seccomp=unconfined --cap-add=IPC_LOCK --cap-add=SYS_NICE --ulimit memlock=-1:-1 --ulimit stack=67108864 --device /dev/infiniband --label io.atlasctl.managed=1 --label io.atlasctl.recipe=qwen3-coder-next-fp8 --rm --name atlas-qwen3-coder-next-fp8 -e HF_HOME=/cache/huggingface -e HF_HUB_OFFLINE=1 -e TRANSFORMERS_OFFLINE=1 -v /home/spark/.cache/huggingface:/cache/huggingface avarok/atlas-gb10:latest spark serve Qwen/Qwen3-Coder-Next-FP8 --port 8888 --host 0.0.0.0 --gpu-memory-utilization 0.92 --max-seq-len 16384 --kv-cache-dtype bf16 --max-batch-size 1 --tool-call-parser qwen3_coder --scheduling-policy slai --oom-guard-mb 1024 --ssm-cache-slots 0

# solo — portable
docker run -d --entrypoint '' --gpus all --ipc=host --shm-size=32gb --network=host --user $(id -u):$(id -g) -v /etc/passwd:/etc/passwd:ro -v /etc/group:/etc/group:ro --security-opt no-new-privileges --security-opt seccomp=unconfined --cap-add=IPC_LOCK --cap-add=SYS_NICE --ulimit memlock=-1:-1 --ulimit stack=67108864 --device /dev/infiniband --label io.atlasctl.managed=1 --label io.atlasctl.recipe=qwen3-coder-next-fp8 --rm --name atlas-qwen3-coder-next-fp8 -e HF_HOME=/cache/huggingface -e HF_HUB_OFFLINE=1 -e TRANSFORMERS_OFFLINE=1 -v $HOME/.cache/huggingface:/cache/huggingface avarok/atlas-gb10:latest spark serve Qwen/Qwen3-Coder-Next-FP8 --port 8888 --host 0.0.0.0 --gpu-memory-utilization 0.92 --max-seq-len 16384 --kv-cache-dtype bf16 --max-batch-size 1 --tool-call-parser qwen3_coder --scheduling-policy slai --oom-guard-mb 1024 --ssm-cache-slots 0

# solo — argv
docker
run
-d
--entrypoint

--gpus
all
--ipc=host
--shm-size=32gb
--network=host
--user
1000:1000
-v
/etc/passwd:/etc/passwd:ro
-v
/etc/group:/etc/group:ro
--security-opt
no-new-privileges
--security-opt
seccomp=unconfined
--cap-add=IPC_LOCK
--cap-add=SYS_NICE
--ulimit
memlock=-1:-1
--ulimit
stack=67108864
--device
/dev/infiniband
--label
io.atlasctl.managed=1
--label
io.atlasctl.recipe=qwen3-coder-next-fp8
--rm
--name
atlas-qwen3-coder-next-fp8
-e
HF_HOME=/cache/huggingface
-e
HF_HUB_OFFLINE=1
-e
TRANSFORMERS_OFFLINE=1
-v
/home/spark/.cache/huggingface:/cache/huggingface
avarok/atlas-gb10:latest
spark
serve
Qwen/Qwen3-Coder-Next-FP8
--port
8888
--host
0.0.0.0
--gpu-memory-utilization
0.92
--max-seq-len
16384
--kv-cache-dtype
bf16
--max-batch-size
1
--tool-call-parser
qwen3_coder
--scheduling-policy
slai
--oom-guard-mb
1024
--ssm-cache-slots
0

