Manage container images on the Bohrium platform. Prefer bohr CLI; fall back to the API for Dockerfile builds and version search.
Since 2023, Bohrium no longer supports VM jobs — container images are required.
json
"bohrium-image": {
"enabled": true,
"apiKey": "YOUR_ACCESS_KEY",
"env": { "ACCESS_KEY": "YOUR_ACCESS_KEY" }
}
bash
# macOS
/bin/bash -c "$(curl -fsSL https://dp-public.oss-cn-beijing.aliyuncs.com/bohrctl/1.0.0/install_bohr_mac_curl.sh)"
# Linux
/bin/bash -c "$(curl -fsSL https://dp-public.oss-cn-beijing.aliyuncs.com/bohrctl/1.0.0/install_bohr_linux_curl.sh)"
source ~/.bashrc && export PATH="$HOME/.bohrium:$PATH"
bash
bohr image list # Custom images (table)
bohr image list --json # JSON
# Public images by type
bohr image list -t "Basic Image"
bohr image list -t "DeePMD-kit"
bohr image list -t "LAMMPS"
bohr image list -t "ABACUS"
bohr image list -t "CP2K"
bohr image list -t "GROMACS"
bohr image list -t "Uni-Mol"
JSON fields: imageId, name, url, status (available/building), creatorName
ScenarioImageCPUregistry.dp.tech/dptech/ubuntu:20.04-py3.10CPU + Intel MPIregistry.dp.tech/dptech/ubuntu:20.04-py3.10-intel2022GPUregistry.dp.tech/dptech/ubuntu:20.04-py3.10-cuda11.6GPU + Intel MPIregistry.dp.tech/dptech/ubuntu:20.04-py3.10-intel2022-cuda11.6
SoftwareImageDeePMD-kitregistry.dp.tech/dptech/deepmd-kit:2.1.5-cuda11.6DPGENregistry.dp.tech/dptech/dpgen:0.10.6LAMMPSregistry.dp.tech/dptech/lammps:29Sep2021GROMACSregistry.dp.tech/dptech/gromacs:2022.2Quantum-Espressoregistry.dp.tech/dptech/quantum-espresso:7.1CP2Kregistry.dp.tech/dptech/cp2k:7.1ABACUSregistry.dp.tech/dptech/abacus:3.0.0
CategorySoftwarePythonpython3.10, pip, Anaconda, Jupyter LabFile toolswget, curl, unzip, rsync, tree, gitEditorsemacs, vimBuild toolscmake, build-essential (GNU)Monitoringhtop, ncdu, net-toolsDP toolsBohrium CLI, DP-Dispatcher, dpdata
VM ImageContainer ImageLBG_DeePMD-kit_2.1.4_v1``registry.dp.tech/dptech/deepmd-kit:2.1.5-cuda11.6``LBG_DP-GEN_0.10.6_v3``registry.dp.tech/dptech/dpgen:0.10.6``LBG_LAMMPS_stable_23Jun2022_v1``registry.dp.tech/dptech/lammps:29Sep2021``gromacs-dp:2020.2``registry.dp.tech/dptech/gromacs:2022.2``LBG_Quantum-Espresso_7.1``registry.dp.tech/dptech/quantum-espresso:7.1``LBG_Common_v1/v2``registry.dp.tech/dptech/ubuntu:20.04-py3.10-cuda11.6``LBG_oneapi_2021_v1``registry.dp.tech/dptech/ubuntu:20.04-py3.10-intel2022-cuda11.6
bash
bohr image pull registry.dp.tech/dptech/deepmd-kit:3.0.0b3-cuda12.1
Requires Docker running locally. Only public and your own custom images are supported.
Replace domain with registry.bohrium.dp.tech (not registry.dp.tech):
bash
docker login registry.bohrium.dp.tech
docker pull registry.bohrium.dp.tech/dptech/ubuntu:22.04-py3.10-intel2022
# Push is not supported
bash
bohr image delete 121510
bohr image delete 121510 121395 # Batch
ScenarioDetailsPublic imagesPersistent cache; no extra pull timeNew custom imagesCache builds in 10-30 min; wait before usingUnused 30 daysCache expires; re-pull on next useWith cacheCPU ~20s boot, GPU ~40s bootWithout cache+10-30 min for image pull
python
import os, requests
AK = os.environ.get("ACCESS_KEY", "")
HEADERS = {"accessKey": AK}
r = requests.get("https://open.bohrium.com/openapi/v2/image/public/version/search",
headers=HEADERS, params={"keyword": "deepmd", "page": 1, "pageSize": 5})
# Returns: {items: [{version, resourceType, size, url, imageName}, ...]}
python
r = requests.get("https://open.bohrium.com/openapi/v2/image/public",
headers=HEADERS, params={"page": 1, "pageSize": 10})
r = requests.get(f"https://open.bohrium.com/openapi/v2/image/public/{image_id}/version",
headers=HEADERS, params={"page": 1, "pageSize": 10})
python
HEADERS_JSON = {**HEADERS, "Content-Type": "application/json"}
# Note: dockerfile field must be base64-encoded
import base64
dockerfile_content = "FROM ubuntu:22.04\nRUN apt-get update && apt-get install -y python3"
dockerfile_b64 = base64.b64encode(dockerfile_content.encode()).decode()
r = requests.post("https://open.bohrium.com/openapi/v2/image/private",
headers=HEADERS_JSON, json={
"name": "my-image", "projectId": 154, "device": "container",
"desc": "Custom training image", "buildType": 1,
"dockerfile": dockerfile_b64,
})
# Validate Dockerfile (also requires base64)
check_b64 = base64.b64encode(b"FROM ubuntu:22.04\nRUN apt-get update").decode()
requests.post("https://open.bohrium.com/openapi/v2/image/dockerfile/check",
headers=HEADERS_JSON, json={"dockerfile": check_b64})
python
# List (must include device and type parameters)
r = requests.get("https://open.bohrium.com/openapi/v2/image/private",
headers=HEADERS, params={"device": "container", "type": "private", "page": 1, "pageSize": 10})
# Returns: {items: [{id, name, url, status, buildType, creatorName, projectName, ...}]}
# Share / unshare
requests.post(f"https://open.bohrium.com/openapi/v2/image/{image_id}/share", headers=HEADERS_JSON)
requests.delete(f"https://open.bohrium.com/openapi/v2/image/{image_id}/share?device=container", headers=HEADERS)
After installing software on a container node, save the environment as a custom image via the Bohrium web UI. Only the system disk is saved; /personal and /share are excluded.
ResourceLimitCustom images10 per project
EndpointVersionReasonGET v1/image/publicv1Caught by /:imageId routeGET v1/image/privatev1SamePOST v2/image/version/addv2Not registeredDELETE v2/image/version/{versionId}v2Not registered
ProblemCauseSolutionbohr image pull failsDocker not runningStart Docker Desktopv2 private param errorMissing required paramsAdd device=container&type=private parametersno permissionNot image creatorCan only manage own imagesv1 /public parse errorRoute conflictUse v2 endpointsWrong image addressUsed name, not full URLMust use registry.dp.tech/dptech/xxx:tagSlow custom image cacheCache takes 10-30 minWait 30 min after buildNo Docker in containerSecurity restrictionUse VM image LBG_Common_v2Create returns decode errdockerfile not base64 encodedUse base64.b64encode(content.encode()).decode()