[sheepdog] [PATCH v3 00/12] A new impelmentation of cluster snapshot

Liu Yuan namei.unix at gmail.com
Tue May 14 10:46:26 CEST 2013


On 05/14/2013 04:31 PM, Kai Zhang wrote:
> *Usage* # save all readonly objects to local path # local path should
> created before this command

We should handle the case when path is not created. Simply call xmkdir
would work.

> # local path could save more than one cluster snapshot collie cluster
> snapshot save /localpath
> 
> 
> # list all cluster snapshot saved in local path collie cluster
> snapshot list /localpath
> 
> 
> # load a snapshot to a cluster # this will format cluster firstly #
> option "-e" is used to specify snapshot if there are more than one

Drop the epoch concept from snapshot object. When we rollback to a
specific snapshot, the epoch should always be 1.

Use tag and idx to index the snapshot as vdi snapshot operations.

> # default is to load the last snapshot if user doesn't specify collie
> cluster snapshot load /localpath
> 

No, don't progress unless tag or idx is specified from user.

> 
> More details of usage and command outputs could be found in tests/030
> which has been modified for new implementation.
> 
> *TODO* 1. compression of snapshot data in sha1 file
>  2. only read snapshot objects created after the latest cluster
> snapshot taken

Current VDI read don't handle sparse volume, so you will store zero
objects? This should be avoided even in the fist implementation.

I'll review the code later.

Thanks,
Yuan



More information about the sheepdog mailing list