[Gluster-devel] [Geo-rep] file in source volume do not cost capacity, but after geo-replication file in target volume does.

[Aravinda]

Geo-replication uses rsync with --inplace option to retain same GFID. 
Rsync will not allow to use both -S and --inplace option.

Tar+SSH mode is sparse aware. (gluster volume geo-replication <MASTER> 
<SLAVEHOST>::<SLAVEVOL> config use_tarssh True)

regards
Aravinda

On 08/08/2016 05:35 PM, Wei-Ming Lin wrote:
> HI ALL,
>
> I create a geo-rep between two volumes, detail as follow:
>
> SRC volume info:
>
> Volume Name: ivan
> Type: Disperse
> Volume ID: 6c4a1a7c-0516-47c4-a298-dabf00dd268c
> Status: Started
> Number of Bricks: 1 x (2 + 1) = 3
> Transport-type: tcp
> Bricks:
> Brick1: giting4:/export/ivan1
> Brick2: giting5:/export/ivan1
> Brick3: giting6:/export/ivan1
> Options Reconfigured:
> changelog.changelog: on
> geo-replication.ignore-pid-check: on
> geo-replication.indexing: on
> performance.readdir-ahead: on
>
> TAR volume info:
>
> Volume Name: tar
> Type: Disperse
> Volume ID: e5147094-bd19-4b41-ae12-f6b84a89a18b
> Status: Started
> Number of Bricks: 1 x (2 + 1) = 3
> Transport-type: tcp
> Bricks:
> Brick1: giting1:/export/ec1/fs
> Brick2: giting2:/export/ec1/fs
> Brick3: giting3:/export/ec1/fs
> Options Reconfigured:
> performance.readdir-ahead: on
> cluster.enable-shared-storage: enable
>
>
> RR status:
> [root at giting4 ivan]# gluster volume geo-replication ivan giting1::tar 
> status
>
> MASTER NODE    MASTER VOL    MASTER BRICK     SLAVE USER  SLAVE       
>     SLAVE NODE    STATUS     CRAWL STATUS LAST_SYNCED
> ------------------------------------------------------------------------------------------------------------------------------------------
> giting4        ivan          /export/ivan1    root  giting1::tar   
>  giting3       Active     Changelog Crawl  2016-08-09 01:14:11
> giting6        ivan          /export/ivan1    root  giting1::tar   
>  giting1       Passive    N/A  N/A
> giting5        ivan          /export/ivan1    root  giting1::tar   
>  giting2       Passive    N/A  N/A
>
> and when I use cmd :
>
> # dd if=/dev/zero of=/mnt/ivan/10G_seek_file bs=1M count=0 seek=10240
>
> to generate a empty file with 10G size at mount point of ivan (source 
> volume),
>
> because "count=0", this file will not cost real capacity,
>
> # du 10G_seek_file -h will get :
>
> 8.0K    10G_seek_file
>
> and I also expect that target volume will have 10G_seek_file that do 
> not cost capacity.
>
> but when I check mount point of tar (target volume),
>
> geo-replication happened,
>
> but 10G_seek_file in target volume does cost capacity.
>
> # du 10G_seek_file -h will get :
>
> 10G  10G_seek_file.
>
> I know that geo-rep in gluster using rsync,
>
> so I using rsync to test the same geo-rep,
>
> found that if I do rsync  with "-S" option,
>
> file in target volume will not cost capacity as source volume.
>
> So, is this a bug?
>
> or is there any configure in gluster can handle this?
>
> thanks.
>
>
>
>
>
>
>
>
>
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at gluster.org
> http://www.gluster.org/mailman/listinfo/gluster-devel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20160809/a7362da9/attachment.html>