处理 Red 状态的 ES 索引 GET _cat/shards?v=true&h=index,shard,prirep,state,node,unassigned.reason&s=state 1 2 ops-pod-loggie-2026.06.11 0 p UNASSIGNED NODE_LEFT ops-pod-loggie-2026.06.11 0 r UNASSIGNED ALLOCATION_FAILED 尝试重新分配 1 POST _cluster/reroute?retry_failed=true 或者直接删除 1 DELETE ops-pod-loggie-2026.06.11 定时清理索引 全部索引仅保留 7 天 创建 ILM 策略 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 PUT _ilm/policy/ops-delete-after-7d { "policy": { "phases": { "hot": { "actions": {} }, "delete": { "min_age": "7d", "actions": { "delete": {} } } } } } 创建索引模板 1 2 3 4 5 6 7 8 9 10 PUT _index_template/ops-delete-after-7d

1. 启动容器 设置镜像 1 export IMAGE=nousresearch/hermes-agent:v2026.6.5 设置目录权限 1 2 mkdir -p hermes-agent-home chmod -R 777 hermes-agent-home 设置环境变量 1 2 3 cat <<EOF > hermes-agent-home/.env GATEWAY_ALLOW_ALL_USERS=true EOF 启动 hermes-agent 1 2 3 4 5 6 7 8 nerdctl run -d \ --name hermes-agent \ --restart always \ --user hermes \ --network host \ -v $(pwd)/hermes-agent-home:/opt/data \ -v $(pwd)/openclaw-home/.openclaw:/opt/data/.openclaw \ $IMAGE gateway run 启动 hermes-agent dashboard 1 2 3 4 5 6 7 nerdctl run -d \ --name hermes-dashboard \ --restart always \ -p 9119:9119 \ -v $(pwd)/hermes-agent-home:/opt/data \ -e GATEWAY_HEALTH_URL=http://${node_IP}:8642 \ $IMAGE dashboard --host 0.0.0.0 --insecure 删除容器 1 nerdctl rm

1. 安装部署 下载部署文件 1 wget https://raw.githubusercontent.com/juicedata/juicefs-csi-driver/master/deploy/k8s.yaml 任意非 master 节点查看 kubelet 的安装目录 1 ps -ef|grep kubelet |grep root-dir 如果有值，则需要执行替换；否则直接安装。 1 sed 's@/var/lib/kubelet@{{KUBELET_DIR}}@g' k8s.yaml 1 kubectl apply -f k8s.yaml 查看 Pod 状态 1 kubectl -n kube-system get pods -l app.kubernetes.io/name=juicefs-csi-driver 2. 创建存储 静态绑定 1 2 3 4 5 6 7 8 9 10 11 12 13 14 apiVersion: v1 metadata: name: juicefs-secret namespace: default labels: juicefs.com/validate-secret: "true" kind: Secret type: Opaque stringData: name: ${JUICEFS_NAME} token: ${JUICEFS_TOKEN} access-key: ${ACCESS_KEY}

1. 架构 1 2 3 4 5 6 7 ┌──────────────────────┐ A2A/JSON-RPC ┌──────────────────────┐ │ OpenClaw 服务器 A │ ◄──────────────────────────► │ OpenClaw 服务器 B │ │ │ (Tailscale / 内网) │ │ │ Agent: AGI │ │ Agent: Coco │ │ A2A

1. 下载 Chart 包 1 git clone https://github.com/loggie-io/installation 2. 安装 loggie 1 cd installation/helm-chart 1 helm install loggie ./ -nloggie --create-namespace 3. 全局配置修改 移除标准输出前面的时间戳 1 kubectl -n loggie edit cm loggie-config-loggie 将 parseStdout 改为 true，移除标准输出前面的时间戳。 1 2 3 4 5 6 config: loggie: discovery: enabled: true kubernetes: parseStdout: true podlogs 改成 varlog 挂载 1 kubectl -n loggie edit ds loggie 1 2 3 4 5 6 7 8 9 volumeMounts: - mountPath: /var/log name: varlog readOnly: true volumes: - hostPath: path: /var/log