[sheepdog-users] is --nohalt dangerous?

[Dietmar Maurer]

So maybe the following would make --nohalt a bit safer?

diff --git a/sheep/group.c b/sheep/group.c
index 059656e..e9e2861 100644
--- a/sheep/group.c
+++ b/sheep/group.c
@@ -1079,6 +1079,9 @@ void sd_leave_handler(struct sd_node *left, struct sd_node
        if (sys_can_halt()) {
                if (current_vnode_info->nr_zones < sys->nr_copies)
                        sys_stat_set(SD_STATUS_HALT);
+       } else {
+               if (current_vnode_info->nr_zones <= (sys->nr_copies/2))
+                       sys_stat_set(SD_STATUS_HALT);
        }
 
        sockfd_cache_del(&left->nid);


> -----Original Message-----
> From: Liu Yuan [mailto:namei.unix at gmail.com]
> Sent: Mittwoch, 18. Juli 2012 08:56
> To: Dietmar Maurer
> Cc: sheepdog-users at lists.wpkg.org
> Subject: Re: [sheepdog-users] is --nohalt dangerous?
> 
> On 07/18/2012 02:27 PM, Dietmar Maurer wrote:
> >> We just provide the safest default to run sheep generally and users
> >> can specify --nohalt options if he don't need it.
> >
> > What do you mean by 'if he don't need it.'? Data consistency is always a
> requirement (else I would store in RAM).
> >
> 
> I mean, users don't need this strict requirement, that we need at least nodes
> >= copies to run the cluster.
> 
> Thanks,
> Yuan
> 
>