automation-suite
2023.10
false
- 概述
- 要求
- 推荐:部署模板
- 手动:准备安装
- 手动:准备安装
- 步骤 1:为离线安装配置符合 OCI 的注册表
- 步骤 2:配置外部对象存储
- 步骤 3:配置 High Availability Add-on
- 步骤 4:配置 Microsoft SQL Server
- 步骤 5:配置负载均衡器
- 步骤 6:配置 DNS
- 步骤 7:配置磁盘
- 步骤 8:配置内核和操作系统级别设置
- 步骤 9:配置节点端口
- 步骤 10:应用其他设置
- 步骤 12:验证并安装所需的 RPM 包
- 步骤 13:生成 cluster_config.json
- 证书配置
- 数据库配置
- 外部对象存储配置
- 预签名 URL 配置
- 符合 OCI 的外部注册表配置
- Disaster Recovery:主动/被动和主动/主动配置
- High Availability Add-on 配置
- 特定于 Orchestrator 的配置
- Insights 特定配置
- Process Mining 特定配置
- Document Understanding 特定配置
- Automation Suite Robot 特定配置
- 监控配置
- 可选:配置代理服务器
- 可选:在多节点 HA 就绪生产集群中启用区域故障恢复
- 可选:传递自定义 resolv.conf
- 可选:提高容错能力
- install-uipath.sh 参数
- 添加具有 GPU 支持的专用代理节点
- 为 Task Mining 添加专用代理节点
- 连接 Task Mining 应用程序
- 为 Automation Suite Robot 添加专用代理节点
- 步骤 15:为离线安装配置临时 Docker 注册表
- 步骤 16:验证安装的先决条件
- 手动:执行安装
- 安装后
- 集群管理
- 监控和警示
- 迁移和升级
- 特定于产品的配置
- 最佳实践和维护
- 故障排除
- 如何在安装过程中对服务进行故障排除
- 如何卸载集群
- 如何清理离线工件以改善磁盘空间
- 如何清除 Redis 数据
- 如何启用 Istio 日志记录
- 如何手动清理日志
- 如何清理存储在 sf-logs 存储桶中的旧日志
- 如何禁用 AI Center 的流日志
- 如何对失败的 Automation Suite 安装进行调试
- 如何在升级后从旧安装程序中删除映像
- 如何禁用 TX 校验和卸载
- 如何从 Automation Suite 2022.10.10 和 2022.4.11 升级到 2023.10.2
- 如何手动将 ArgoCD 日志级别设置为 Info
- 如何扩展 AI Center 存储
- 如何为外部注册表生成已编码的 pull_secret_value
- 如何解决 TLS 1.2 中的弱密码问题
- 单节点升级在结构阶段失败
- 从 2021.10 自动升级后,集群运行状况不佳
- 由于 Ceph 运行状况不佳,升级失败
- 由于空间问题,RKE2 未启动
- 卷无法装载,且仍处于附加/分离循环状态
- 由于 Orchestrator 数据库中的传统对象,升级失败
- 并行升级后,发现 Ceph 集群处于降级状态
- Insights 组件运行状况不佳导致迁移失败
- Apps 服务升级失败
- 就地升级超时
- Docker 注册表迁移卡在 PVC 删除阶段
- 升级到 2023.10 或更高版本后 AI Center 配置失败
- 在离线环境中升级失败
- 升级期间 SQL 验证失败
- 快照-控制器-crds Pod 在升级后处于 CrashLoopBackOff 状态
- Longhorn REST API 端点升级/重新安装错误
- 运行诊断工具
- 使用 Automation Suite 支持捆绑包
- 探索日志
Pod 在 Init:0/X 中卡住
Linux 版 Automation Suite 安装指南
Last updated 2024年12月3日
Pod 在 Init:0/X 中卡住
使用 LH 卷的“说明 Pod”卡在 Init:0/X(其中 X 是表示容器数量的整数),并且 Pod 上的“kubectl describe”命令针对事件中的卷返回“MapVolume.MapPodDevice failed”。
要解决此问题,请运行以下命令:
for podPv in $(kubectl get events -A -o json | jq -r '.items[] | select(.reason == "FailedMapVolume" and .involvedObject.kind == "Pod" and (.message | contains("MapVolume.MapPodDevice failed for volume") and contains("Could not mount \"/dev/longhorn"))) | .involvedObject.namespace + "=" + .involvedObject.name + "=" + (.message | match("(pvc-[0-9a-z-]+)").captures[0].string )') ; do
echo "Found 'FailedMapVolume' error: '${podPv}'"
NS=$(echo "${podPv}" | cut -d'=' -f1)
POD=$(echo "${podPv}" | cut -d'=' -f2)
PV=$(echo "${podPv}" | cut -d'=' -f3)
[[ -z "${NS}" ]] && echo "Could not extract namespace for error: '${podPv}'" && continue
[[ -z "${POD}" ]] && echo "Could not extract pod name for error: '${podPv}'" && continue
[[ -z "${PV}" ]] && echo "Could not extract Persistent Volume for error: '${podPv}'" && continue
controller_data=$(kubectl -n "${NS}" get po "${POD}" -o json | jq -r '[.metadata.ownerReferences[] | select(.controller==true)][0] | .kind + "=" + .name')
[[ -z "$controller_data" ]] && echo "Could not determine owner for pod: ${POD} in namespace: ${NS}" && continue
CONTROLLER_KIND=$(echo "${controller_data}" | cut -d'=' -f1)
[[ -z "${CONTROLLER_KIND}" ]] && echo "Could not extract controller kind for pod: '${POD}' in namespace: '${NS}' && continue
CONTROLLER_NAME=$(echo "${controller_data}" | cut -d'=' -f2)
[[ -z "${CONTROLLER_NAME}" ]] && echo "Could not extract controller name for pod: '${POD}' in namespace: '${NS}' && continue
if [[ $CONTROLLER_KIND == "ReplicaSet" ]]
then
controller_data=$(kubectl -n "${NS}" get "${CONTROLLER_KIND}" "${CONTROLLER_NAME}" -o json | jq -r '[.metadata.ownerReferences[] | select(.controller==true)][0] | .kind + "=" + .name')
CONTROLLER_KIND=$(echo "${controller_data}" | cut -d'=' -f1)
[[ -z "${CONTROLLER_KIND}" ]] && echo "Could not extract controller kind(from rs) for pod: '${POD}' in namespace: '${NS}'" && continue
CONTROLLER_NAME=$(echo "${controller_data}" | cut -d'=' -f2)
[[ -z "${CONTROLLER_NAME}" ]] && echo "Could not extract controller name(from rs) for pod: '${POD}' in namespace: '${NS}'" && continue
fi
org_replicas=$(kubectl -n "${NS}" get "$CONTROLLER_KIND" "$CONTROLLER_NAME" -o json | jq -r '.status.replicas')
echo "Scaling down ${CONTROLLER_KIND}/${CONTROLLER_NAME}"
kubectl -n "${NS}" patch "$CONTROLLER_KIND" "${CONTROLLER_NAME}" -p "{\"spec\":{\"replicas\":0}}"
if kubectl -n "${NS}" get pod "${POD}" ; then
kubectl -n "${NS}" wait --for=delete pod "${POD}" --timeout=300s
fi
if kubectl get volumeattachment | grep -q "${PV}"; then
volumeattachment_id=$(kubectl get volumeattachment | grep "${PV}" | awk '{print $1}')
kubectl delete volumeattachment ${volumeattachment_id}
fi
[[ -z "$org_replicas" || "${org_replicas}" -eq 0 ]] && org_replicas=1
echo "Scaling up ${CONTROLLER_KIND}/${CONTROLLER_NAME}"
kubectl -n "${NS}" patch "$CONTROLLER_KIND" "${CONTROLLER_NAME}" -p "{\"spec\":{\"replicas\":${org_replicas}}}"
done
for podPv in $(kubectl get events -A -o json | jq -r '.items[] | select(.reason == "FailedMapVolume" and .involvedObject.kind == "Pod" and (.message | contains("MapVolume.MapPodDevice failed for volume") and contains("Could not mount \"/dev/longhorn"))) | .involvedObject.namespace + "=" + .involvedObject.name + "=" + (.message | match("(pvc-[0-9a-z-]+)").captures[0].string )') ; do
echo "Found 'FailedMapVolume' error: '${podPv}'"
NS=$(echo "${podPv}" | cut -d'=' -f1)
POD=$(echo "${podPv}" | cut -d'=' -f2)
PV=$(echo "${podPv}" | cut -d'=' -f3)
[[ -z "${NS}" ]] && echo "Could not extract namespace for error: '${podPv}'" && continue
[[ -z "${POD}" ]] && echo "Could not extract pod name for error: '${podPv}'" && continue
[[ -z "${PV}" ]] && echo "Could not extract Persistent Volume for error: '${podPv}'" && continue
controller_data=$(kubectl -n "${NS}" get po "${POD}" -o json | jq -r '[.metadata.ownerReferences[] | select(.controller==true)][0] | .kind + "=" + .name')
[[ -z "$controller_data" ]] && echo "Could not determine owner for pod: ${POD} in namespace: ${NS}" && continue
CONTROLLER_KIND=$(echo "${controller_data}" | cut -d'=' -f1)
[[ -z "${CONTROLLER_KIND}" ]] && echo "Could not extract controller kind for pod: '${POD}' in namespace: '${NS}' && continue
CONTROLLER_NAME=$(echo "${controller_data}" | cut -d'=' -f2)
[[ -z "${CONTROLLER_NAME}" ]] && echo "Could not extract controller name for pod: '${POD}' in namespace: '${NS}' && continue
if [[ $CONTROLLER_KIND == "ReplicaSet" ]]
then
controller_data=$(kubectl -n "${NS}" get "${CONTROLLER_KIND}" "${CONTROLLER_NAME}" -o json | jq -r '[.metadata.ownerReferences[] | select(.controller==true)][0] | .kind + "=" + .name')
CONTROLLER_KIND=$(echo "${controller_data}" | cut -d'=' -f1)
[[ -z "${CONTROLLER_KIND}" ]] && echo "Could not extract controller kind(from rs) for pod: '${POD}' in namespace: '${NS}'" && continue
CONTROLLER_NAME=$(echo "${controller_data}" | cut -d'=' -f2)
[[ -z "${CONTROLLER_NAME}" ]] && echo "Could not extract controller name(from rs) for pod: '${POD}' in namespace: '${NS}'" && continue
fi
org_replicas=$(kubectl -n "${NS}" get "$CONTROLLER_KIND" "$CONTROLLER_NAME" -o json | jq -r '.status.replicas')
echo "Scaling down ${CONTROLLER_KIND}/${CONTROLLER_NAME}"
kubectl -n "${NS}" patch "$CONTROLLER_KIND" "${CONTROLLER_NAME}" -p "{\"spec\":{\"replicas\":0}}"
if kubectl -n "${NS}" get pod "${POD}" ; then
kubectl -n "${NS}" wait --for=delete pod "${POD}" --timeout=300s
fi
if kubectl get volumeattachment | grep -q "${PV}"; then
volumeattachment_id=$(kubectl get volumeattachment | grep "${PV}" | awk '{print $1}')
kubectl delete volumeattachment ${volumeattachment_id}
fi
[[ -z "$org_replicas" || "${org_replicas}" -eq 0 ]] && org_replicas=1
echo "Scaling up ${CONTROLLER_KIND}/${CONTROLLER_NAME}"
kubectl -n "${NS}" patch "$CONTROLLER_KIND" "${CONTROLLER_NAME}" -p "{\"spec\":{\"replicas\":${org_replicas}}}"
done