虚拟机编辑添加新磁盘超时后，导致所有vmdk文件丢失

[SGMY]

环境：VMware vShpere vsan6.7
现象：对正常在用的虚拟机编辑，新添加一块磁盘时，确定后任务进度卡在百分之十几不动，直至添加失败出错"与远程主机通信时出错"，然后该虚拟机大小就变成只有几十兆大小了，导航至vsan后端到该虚拟机目录查看，发现所有的vmdk文件都没有了，已经出现好几次这种问题了，是否有解决方案

[TheBobkin]

@茂勇 上官

 

Please note this is an English-language sub-community - if you want to ask questions in your native language then please use the correct sub-community:

https://community.broadcom.com/groups/communities/communityhomeblogs?CommunityKey=ba6cb483-29c6-45e9-a764-afebe9f1e4a9

 

Using Google translate:

 

"Environment: VMware vShpere vsan6.7

Phenomenon: When editing a virtual machine in normal use and adding a new disk, the task progress is stuck at more than 10% after confirmation, until the addition fails and an error "Error in communicating with the remote host" occurs. Then the size of the virtual machine becomes only tens of megabytes. Navigate to the vsan backend to check the virtual machine directory and find that all vmdk files are gone. This problem has occurred several times. Is there a solution?"

 

From your screenshot, the vmdk descriptors and also the .vmx appear to be gone - you should check the shell.log (on all hosts) and hostd.log to check whether someone/something is deleting these, that is not normal and not something vSphere/ESXi/vSAN would do under any circumstances.

It is also possible that deletion of these occurred some time in the past and only when you go to make some other change that triggers update/refresh of these files (and/or release of locks e.g. power-off of VM) that deletion of these is able to complete.

It is possible that whatever/whomever deleted the data only deleted the descriptor data (e.g. only the .vmdk that points to the data object, not the data object itself e.g. via 'rm' or another command) - you can check whether the actual data object backing the vmdk is still present by looking at the data derived from 'esxcli vsan debug object list --all > /tmp/objout123' (note this command may not work with the '--all' flag in older builds).