当前位置：首页 > news >正文

77. Rancher 2.11 到 2.12 升级清单——修复审计日志、集群和界面缓存

news 2026/3/27 5:09:41

Environment 环境

From: Rancher 2.11.x
来自：牧场主 2.11.x
To: Rancher 2.12..x
收件人：牧场主 2.12..x
Kubernetes: 1.31 - 1.33 (see Support Matrix)
Kubernetes：1.31 - 1.33（参见支持矩阵）
Install type: RKE2
安装类型：RKE2

Procedure 程序

Goal: Upgrade Rancher 2.11 → 2.12without losing logs or clusters. This checklist covers common issues and settings to check
目标：在不丢失日志或集群的情况下，将 Rancher 2.11 升级→2.12。本清单涵盖了常见问题和需要检查的设置

Upgrade Checklist 升级清单

1. Check Your Setup
1. 检查你的设备

Run the official Upgrade Checklist and Support Matrix.
运行官方升级清单和支持矩阵。

Example: verify version helm list -n cattle-system | grep rancher

2. If you need to Keep Audit Logs ON after upgrading
2. 升级后是否需要保持审计日志开启状态

Audit logs turn OFF after upgrade.
升级后审计日志会关闭。
Add these flags to helm upgrade:
在升级舵手时添加以下旗帜：

--set auditLog.enabled=true --set auditLog.level=2

Level 2 = default (who + what). Use Level 3 only if needed.
2级=默认（谁 + 什么）。只有在需要时才使用3级。

Full example: 完整示例：

helm upgrade rancher rancher-stable/rancher \ --namespace cattle-system \ --set hostname=YOUR_HOST \ --set auditLog.enabled=true \ --set auditLog.level=2 \ --version 2.12.1

3. Fix Old Downstream Clusters
3. 修复旧的下游集群

Imported clusters may disconnect (GitHub Issue #51066).
导入的集群可能会断开连接（GitHub issue #51066）。

Step A – Find clusters
步骤 A——寻找集群

Execute the following commands.
执行以下命令。
Extract the Downstream cluster names present in both results.
提取两个结果中存在的下游集群名称。

kubectl get clusters.provisioning.cattle.io -A kubectl get clusters.management.cattle.io

Step B – Check fields (replace )
步骤 B——检查字段（替换）

Check theclusters.management.cattle.ioresource.
查看clusters.management.cattle.io资源。
If all of the following items are not set to"", that cluster is affected.
如果以下所有项目都未被设置为“”，该集群会受到影响。

kubectl get clusters.management.cattle.io <name> -o jsonpath='{.spec.desiredAgentImage}{.spec.desiredAuthImage}{.spec.clusterSecrets.privateRegistryURL}'

Step C – Clear fields
步骤 C——清除场

# Backup first kubectl get clusters.management.cattle.io <name> -o yaml > backup-<name>.yaml # Edit – set all three to "" kubectl edit clusters.management.cattle.io <name>

4. Check UI Cache Size
4. 检查界面缓存大小

New cache speeds UI but uses disk. UI Server-Side Pagination docs
新缓存会加快界面速度，但使用磁盘。UI 服务器端分页文档

# Local kubectl exec -n cattle-system deploy/rancher -- ls -lh /var/lib/rancher | grep informer_object_cache # Downstream kubectl exec -n cattle-system deploy/cattle-cluster-agent -- ls -lh /var/lib/rancher | grep informer_object_cache

Example output: 示例输出：

In this example, cache size is 18M + 32K + 6.2M = 25MBytes.
在这个例子中，缓存大小为 18M + 32K + 6.2M = 25MBytes。

-rw------- 1 root jail-accessors 18M Nov 12 06:12 informer_object_cache.db -rw------- 1 root jail-accessors 32K Nov 12 06:12 informer_object_cache.db-shm -rw------- 1 root jail-accessors 6.2M Nov 12 06:12 informer_object_cache.db-wal

Warning:If > 1GB and growing fast, adjust storage or disable cache.
警告：如果> 1GB，且增长很快，建议调整存储或禁用缓存。

5. Adjust or Disable Cache
5. 调整或禁用缓存

Set the size of ephemeral storage or disable the cache to prevent the cache from exceeding ephemeral storage.
设置临时存储大小或禁用缓存以防止缓存超过临时存储。

Set ephemeral-storage =2 × biggest etcd snapshot. (prevents pod crashes)
设置 ephemeral-storage =2 最大× etcd 快照。（防止舱体坠毁）
Turn OFF if issues: Global → Feature Flags → ui-sql-cache →False.
关闭OF：U-SQL-CACHE → 全局 → 功能标志 →False。