[sheepdog] [PATCH] dog: allow vdi snapshot without snapshot tag

Hitoshi Mitake mitake.hitoshi at lab.ntt.co.jp
Fri Sep 19 03:36:24 CEST 2014


At Thu, 18 Sep 2014 18:47:24 +0800,
Ruoyu wrote:
> 
> 
> On 2014年09月18日 14:54, Hitoshi Mitake wrote:
> > The commit a21bf27906b23448b92cca9943e1019105ffac2f makes
> > $ dog vdi snapshot <vdi>
> > fail if the <vdi> is an ordinal vdi because newly created VDIs have
> > snapid 0. This patch avoids the failure with checking the VDI is
> > snapshot or not.
> >
> > Cc: Ruoyu <liangry at ucweb.com>
> > Signed-off-by: Hitoshi Mitake <mitake.hitoshi at lab.ntt.co.jp>
> Thanks Hitoshi.
> The original purpose is to forbid duplicated snapshot tag.
> But I found it is not implemented after applying your patch.
> 
> $ dog/dog vdi snapshot test
> liangry at linuxmint ~/github/sheepdog.orig $ dog/dog vdi list
> Name Id Size Used Shared Creation time VDI id Copies Tag
> s test 2 4.0 MB 0.0 MB 4.0 MB 2014-09-10 14:46 7c2b26 3
> s test 3 4.0 MB 0.0 MB 4.0 MB 2014-09-10 15:30 7c2b27 3
> s test 4 4.0 MB 0.0 MB 4.0 MB 2014-09-18 18:41 7c2b28 3
> s test 5 4.0 MB 0.0 MB 4.0 MB 2014-09-18 18:41 7c2b29 3
> test 0 4.0 MB 0.0 MB 4.0 MB 2014-09-18 18:43 7c2b2a 3

Thanks for your testing and sorry for my mistake, I should use
dog_read_object() in a case of existing VDI instead of read_vdi_obj().

I'll send v2 soon.

Thanks,
Hitoshi

> 
> > ---
> >   dog/vdi.c | 14 ++++++++++----
> >   1 file changed, 10 insertions(+), 4 deletions(-)
> >
> > diff --git a/dog/vdi.c b/dog/vdi.c
> > index fa6130e..0a8a2e9 100644
> > --- a/dog/vdi.c
> > +++ b/dog/vdi.c
> > @@ -567,6 +567,7 @@ static int vdi_snapshot(int argc, char **argv)
> >   	int vs_count = 0;
> >   	struct node_id owners[SD_MAX_COPIES];
> >   	int nr_owners = 0, nr_issued_prevent_inode_update = 0;
> > +	bool fail_if_snapshot = false;
> >   
> >   	if (vdi_cmd_data.snapshot_id != 0) {
> >   		sd_err("Please specify a non-integer value for "
> > @@ -584,6 +585,15 @@ static int vdi_snapshot(int argc, char **argv)
> >   	case SD_RES_NO_TAG:
> >   		break;
> >   	default:
> > +		fail_if_snapshot = true;
> > +		break;
> > +	}
> > +
> > +	ret = read_vdi_obj(vdiname, 0, "", &vid, inode, SD_INODE_HEADER_SIZE);
> > +	if (ret != EXIT_SUCCESS)
> > +		return ret;
> > +
> > +	if (fail_if_snapshot && vdi_is_snapshot(inode)) {
> >   		sd_err("Failed to create snapshot for %s, maybe "
> >   			"snapshot id (%d) or tag (%s) is existed",
> >   			vdiname, vdi_cmd_data.snapshot_id,
> > @@ -591,10 +601,6 @@ static int vdi_snapshot(int argc, char **argv)
> >   		return EXIT_FAILURE;
> >   	}
> >   
> > -	ret = read_vdi_obj(vdiname, 0, "", &vid, inode, SD_INODE_HEADER_SIZE);
> > -	if (ret != EXIT_SUCCESS)
> > -		return ret;
> > -
> >   	if (inode->store_policy) {
> >   		sd_err("creating a snapshot of hypervolume is not supported");
> >   		return EXIT_FAILURE;
> 
> 



More information about the sheepdog mailing list