[sheepdog] [sheepdog-users] Impossible to run vdi check with 1 node

[MORITA Kazutaka]

At Thu, 24 Jan 2013 18:55:02 +0800,
Liu Yuan wrote:
> 
> On 01/24/2013 06:22 PM, MORITA Kazutaka wrote:
> > Panic is not appropriate here.  We should exit with an user friendly
> > message.
> 
> Makes sense, there is the patch
> 
> From b2ba8f9ef3d0929ff001f4ae1e2c60625880eeeb Mon Sep 17 00:00:00 2001
> From: Liu Yuan <tailai.ly at taobao.com>
> Date: Thu, 24 Jan 2013 18:53:58 +0800
> Subject: [PATCH] collie: give more user friendly error message for 'vdi
>  check'
> 
> Signed-off-by: Liu Yuan <tailai.ly at taobao.com>
> ---
>  collie/vdi.c |    4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/collie/vdi.c b/collie/vdi.c
> index b5af7a9..5f2a889 100644
> --- a/collie/vdi.c
> +++ b/collie/vdi.c
> @@ -1506,6 +1506,10 @@ static int vdi_check(int argc, char **argv)
>  		fprintf(stderr, "FATAL: no inode objects\n");
>  		goto out;
>  	}
> +	if (sd_nodes_nr < inode->nr_copies){
> +		fprintf(stderr, "FATAL: do not have enough nodes\n");

FATAL should be used when something serious problem happens on
Sheepdog cluster.  I think this is a wrong usage of 'collie vdi
check', so how about notifying the user that there is not enough nodes
and objects are not fully replicated obviously?

Thanks,

Kazutaka

> +		return EXIT_FAILURE;
> +	}
>  
>  	do_check_repair(vid_to_vdi_oid(vid), inode->nr_copies);
>  	total = inode->vdi_size;
> -- 
> 1.7.9.5
>