[sheepdog] [PATCH] sheep: fix local_get_node_list before the node joined a cluster

Liu Yuan namei.unix at gmail.com
Fri Jun 1 08:11:46 CEST 2012


On 06/01/2012 02:00 PM, Christoph Hellwig wrote:

> On Thu, May 31, 2012 at 11:22:07PM +0800, Liu Yuan wrote:
>>> I wondered about that, too.  The only use case I could find for  the
>>> force flag would be for troubleshooting in case a node didn't join the
>>> cluster.
>>>
>>
>>
>> I think collie cluster info could serve this purpose. So let's make it
>> as .force = 0.
> 
> It turns out collie cluster info actually ends up calling the get noee
> list command, as do about half of the collie sub commands (grep for
> SUBCMD_FLAG_NEED_NODELIST) and they all fail no on a cluster that is
> down.
> 
> This also breaks a lot of testing, so I internally had to revert to the
> older version of the fix.  I think we should move to it for upstream as
> well.
> 


Okay, would you cook a new patch for it? The old fix was drown in my
box, hard to find it now.

Thanks,
Yuan



More information about the sheepdog mailing list