1. 设置环境变量 1 2 export NAMESPACE=xxx export PVC=xxx 2. Dataset 无法就绪 2.1 Fluid 组件问题 1 kubectl -n fluid-system get pod -o wide | grep -v "Running" 可能出现没有正常启动的情况。 2.2 有异常的 Dataset 异常的资源可能导致 Fluid 资源不断重启，需要人工介入删除。 2.3 检查 Worker \ Fuse 副本 worker 副本 1 kubectl -n ${NAMESPACE} get sts -l release=${PVC} 1 kubectl -n ${NAMESPACE} get pod -l release=${PVC},role=juicefs-worker fuse 副本 1 kubectl -n kas-job get ds -l

1. 分析 Fluid 挂载 NFS 存储 查看 Fuse Pod 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 kubectl get pod nfs-demo-fuse-f9wg8 -oyaml apiVersion: v1 kind: Pod metadata: generateName: nfs-demo-fuse- spec: containers: - command: - /usr/local/bin/entrypoint.sh env: - name: FLUID_RUNTIME_TYPE value: thin - name: FLUID_RUNTIME_NS value: default - name: FLUID_RUNTIME_NAME value: nfs-demo - name: MOUNT_POINT value: /runtime-mnt/thin/default/nfs-demo/thin-fuse - name: MOUNT_OPTIONS value: ro image: fluidcloudnative/nfs:v0.1 imagePullPolicy: IfNotPresent lifecycle: preStop: exec: command: - sh - -c - umount /runtime-mnt/thin/default/nfs-demo/thin-fuse name: thin-fuse securityContext: privileged: true volumeMounts:

1. 打包 Fluid Runtime 镜像 创建 fluid_config_init.py 脚本 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 #!/usr/bin/env python import json rawStr = "" with open("/etc/fluid/config.json", "r") as f: rawStr = f.readlines() rawStr = rawStr[0] script = """ #!/bin/sh set -ex MNT_FROM=$mountPoint MNT_TO=$targetPath trap "umount ${MNT_TO}" SIGTERM mkdir -p ${MNT_TO} mount -t lustre -o relatime,flock ${MNT_FROM} ${MNT_TO} sleep inf """ obj = json.loads(rawStr) with open("mount-lustre.sh", "w") as f: f.write('mountPoint="%s"\n' % obj["mounts"][0]["mountPoint"]) f.write('targetPath="%s"\n' % obj["targetPath"]) f.write(script) 只需调整一下 mount 命令即可。 创建启动脚本 entrypoint.sh 1 2 3