Supermicro X9DR3-F服务器RAID-50阵列双故障磁盘更换操作求助
Hey Jesper, let's work through this step by step—dealing with two failed drives in a RAID-50 array sounds stressful, but we can get this sorted if we follow the right process. First, let's clear up the key confusion points:
关于直接插拔更换磁盘的疑问
You can't just yank both failed drives at once and pop in new ones—RAID-50 has specific fault tolerance limits: it’s a combination of RAID-5 (each sub-group can handle one drive failure) and RAID-0. So we need to first confirm which RAID-5 sub-group each failed drive belongs to:
- If the two failed drives are in different sub-groups: Your hot spare should have already kicked in to replace one of them, leaving one sub-group in a degraded state. We can replace the failed drives one at a time, waiting for each rebuild to finish before moving to the next.
- If both failed drives are in the same RAID-5 sub-group: Your array is in a critical state (since RAID-5 can’t handle two failures in one sub-group). Here, you must replace one drive first, wait for the hot spare to rebuild or the new drive to sync, then replace the second one—doing this out of order could crash the array entirely.
正确的操作步骤
First, forget about the Intel RST utility in your Windows VM—it won’t work here! The RAID array is managed by the physical RAID card in your Supermicro server, and your Windows VM only sees the virtual disk presented by ESXi. You need to manage the RAID at the physical host level:
1. 定位故障磁盘及其所属子组
- 重启物理服务器，进入RAID卡的配置工具（大多数搭载LSI MegaRAID卡的X9DR3-F机型，开机时按
Ctrl+R即可进入，具体以屏幕提示为准）。 - 如果你配置了Supermicro的IPMI远程管理，也可以直接登录IPMI网页界面，进入存储或RAID管理模块，查看磁盘状态、故障位置以及每个故障磁盘所属的子组。
2. 更换第一块故障磁盘
- 你的服务器支持热插拔（X9DR3-F的硬盘托架默认支持），找到带琥珀色故障灯的磁盘，安全拔出。
- 插入新磁盘：确保新磁盘容量不小于原磁盘（更小的磁盘无法加入阵列），接口（SAS/SATA）和转速尽量与原磁盘匹配，保障性能一致性。
- 返回RAID配置工具/IPMI界面：
- 如果热备盘已经替换了一块故障盘，需要将新磁盘设置为剩余故障盘的替换盘，或者直接设为新的热备盘，触发重建流程（部分RAID卡会自动识别并启动重建）。
- 如果阵列处于临界状态（两块故障盘同属一个子组），需要先初始化新磁盘并将其加入阵列，等待重建完成（根据磁盘大小，这个过程可能需要数小时）。
3. 更换第二块故障磁盘
- 等第一块磁盘的重建完成、阵列脱离临界状态后，重复上述步骤更换第二块故障磁盘：拔出故障盘，插入新盘，确认系统识别后将其加入阵列或设为热备盘。
关键注意事项
- 先备份数据：即使RAID有冗余，双故障场景仍存在风险，操作前优先将关键数据备份到外部存储设备。
- 不要中断重建：RAID卡重建过程中，避免重启服务器，否则可能导致数据损坏或大幅延长重建时间。
- 使用规格匹配的磁盘：虽然可以用更大容量的磁盘，但超出的部分无法被阵列利用，尽量选择同型号、同转速的磁盘保障稳定性。
备注：内容来源于stack exchange，提问作者Jesper Ekelund
