集群雲盤報錯問題排查

當存儲和計算分離之后，給容器掛載的硬盤一般是塊設備，但這個塊設備可能是在其他主機上，使用iscsi協議來讀寫。

最近遇到一個問題，當iscsi的對應target出問題之后，容器中寫盤出現io error，

 [1808657.505206] blk_update_request: critical medium error, dev sdh, sector 4456448
 [1808657.505695] Buffer I/O error on dev sdh, logical block 557056, lost sync page write
 [1808657.506178] JBD2: Error -5 detected when updating journal superblock for sdh-8.
 [1808657.506657] Aborting journal on device sdh-8.
 [1808657.508173] blk_update_request: critical medium error, dev sdh, sector 4456448
 [1808657.508657] Buffer I/O error on dev sdh, logical block 557056, lost sync page write
 [1808657.509152] JBD2: Error -5 detected when updating journal superblock for sdh-8.

[1808657.510929] blk_update_request: critical medium error, dev sdh, sector 0
[1808657.511396] Buffer I/O error on dev sdh, logical block 0, lost sync page write
[1808657.511875] EXT4-fs error (device sdh): ext4_put_super:800: Couldn't clean up the journal
[1808657.512373] EXT4-fs (sdh): Remounting filesystem read-only
[1808657.512869] EXT4-fs (sdh): previous I/O error to superblock detected
[1808657.514869] blk_update_request: critical medium error, dev sdh, sector 0

 [1808657.515402] Buffer I/O error on dev sdh, logical block 0, lost sync page write

[1808657.516635] VFS: Dirty inode writeback failed for block device sdh (err=-5).

 

查這類問題的時候，因為硬盤不在本機，所以需要根據四種持久化命名方案： by-label、 by-uuid、 by-id 和 by-path等找到對應的硬盤提供側的宿主機，

然后查看對應的狀態即可。