[Openstack-devel] Bug#694518: Bug#694518: [RFR] templates://sheepdog/{sheepdog.templates}

[YunQiang Su]

On Tue, Dec 4, 2012 at 6:53 PM, Justin B Rye <jbr at edlug.org.uk> wrote:

> YunQiang Su wrote:
> > Justin B Rye <jbr at edlug.org.uk> wrote:
> >> Shouldn't this start with something like:
> >>
> >>       Please choose the commandline arguments that should be passed to
> the
> >>       sheepdog daemon. If no argument is given, the default behavior is
> to
> >>       start with the corosync driver on port 7000.
> >
> > In the sheep(8), this is descripte as
> >
> > This option specifies the IP port over which sheep looks for other
> sheepdog
> > cluster nodes.
>
> I'm sorry, I still don't understand.  Yes, when it says "port", it
> means "TCP/IP port"; that was already clear from the usage summary in
> the template.  And the thing sheepdog does is coordinate with other
> nodes in a networked cluster.  But as I went on to ask:
>
> >>    (Does that mean "with 7000 as the corosync driver port" or is it two
> >>    things, "start on port 7000, using the corosync driver"?  Or will
> >>    sheepdog users find it obvious?)
>
This port is for server Qemu.
This port is nothing matter with corosync

>
> Unfortunately the manual doesn't explain what the "corosync driver" is
> or how it is involved in this; it mentions that "KVM/QEMU virtual
> machines use the sheep daemon via a block driver available in
> qemu(1)", and there's an option "-c, --cluster - specify the cluster
> driver", but are those talking about the same thing?  If I answered
> this template's question with "-p 7777 -c custom", does that mean
> QEMU would be using the custom driver instead of the corosync one?
>
There are several software to maintain cluster, such as zookeeper etc.
These softwares is used to keep service on, for example it can switch to
another server when the main server is done.
sheepdog now can work with corosync and zookeeper. Here we cannot
config the way sheepdog work with corosync and zookeeper.

>
> If it's likely that potential users of this package will understand
> this by the time they encounter the template, then the ambiguity may
> not matter, but I'm checking because it isn't clear to me.
>
This description can be better, don't confuse the port with corosync.

> --
> JBR     with qualifications in linguistics, experience as a Debian
>         sysadmin, and probably no clue about this particular package
>



-- 
YunQiang Su
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.alioth.debian.org/pipermail/openstack-devel/attachments/20121205/e073cd58/attachment.html>