Wed, 7 Nov 2001 15:06:18 -0000

------_=_NextPart_001_01C1679D.C098CB90
Content-Type: text/plain

Mmmm.  I don't quite understand what you are getting at.  Do you mean a node
has failed in addition to having failed on giving the magic weighting.  You
mean two failures?  Chances?  About the same as saying I'm node "some other"
node during a failure. 

Cheers, 
     Mark. 

-----Original Message----- 
From:		Alan Cox [SMTP:alan@lxorguk.ukuu.org.uk] 
Sent:		Wednesday 07 November 2001 14:27 
To:		Mark@MarkAndrewSmith.co.uk 
Cc:		alan@lxorguk.ukuu.org.uk; lmb@suse.de;
linux-ha-dev@lists.community.tummy.com; linux-ha@muc.de 
Subject:	Re:  

> Surely it would be just a matter of setting a setting to 1 if a node is 
> attached to an atomic clock via a mesh and 0 if not.  Then  
in the voting 
> described below, A and B win and C to G loose by this overriding 
> consideration.  Upon loosing a vote, should they  
snap into alignment 
> straight away?  Perhaps this should be soft coded too to prevent data 
> problems.....?  

Now suppose a node  
fails by emitting the wrong 0 or 1 magic weighting. It 
gets really horrible because you are using a cluster resource to control 
cluster  
membership. 

------_=_NextPart_001_01C1679D.C098CB90
Content-Type: text/html
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV=3D"Content-Type" CONTENT=3D"text/html; =
charset=3Dus-ascii">
<META NAME=3D"Generator" CONTENT=3D"MS Exchange Server version =
5.5.2653.12">
<TITLE>RE: </TITLE>
</HEAD>
<BODY>

&nbsp;
 Mmmm.&nbsp; I don't quite understand what you are =
getting at.&nbsp; Do you mean a node has failed in addition to having =
failed on giving the magic weighting.&nbsp; You mean two =
failures?&nbsp; Chances?&nbsp; About the same as saying I'm node =
&quot;some other&quot; node during a failure.

&nbsp;
 Cheers, 
 &nbsp;&nbsp;&nbsp;&nbsp; Mark. 
 &nbsp;
 -----Original Message----- 
 From:&nbsp;&nbsp; =
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Alan Cox =
[SMTP:alan@lxorguk.ukuu.org.uk] 
 Sent:&nbsp;&nbsp; =
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Wednesday 07 November 2001 =
14:27 
 To:&nbsp;&nbsp;&nbsp;&nbsp; =
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; Mark@MarkAndrewSmith.co.uk =

 Cc:&nbsp;&nbsp;&nbsp;&nbsp; =
&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; alan@lxorguk.ukuu.org.uk; =
lmb@suse.de; linux-ha-dev@lists.community.tummy.com; linux-ha@muc.de =

 Subject:&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp; =
Re:&nbsp; 
 &nbsp;
 &gt; Surely it would be just a matter of setting a =
setting to 1 if a node is 
 &gt; attached to an atomic clock via a mesh and 0 if =
not.&nbsp; Then&nbsp; 
 in the voting 
 &gt; described below, A and B win and C to G loose =
by this overriding 
 &gt; consideration.&nbsp; Upon loosing a vote, =
should they&nbsp; 
 snap into alignment 
 &gt; straight away?&nbsp; Perhaps this should be =
soft coded too to prevent data 
 &gt; problems.....?&nbsp; 
 &nbsp;
 Now suppose a node&nbsp; 
 fails by emitting the wrong 0 or 1 magic weighting. =
It 
 gets really horrible because you are using a cluster =
resource to control 
 cluster&nbsp; 
 membership. 
 &nbsp;
 &nbsp;

</BODY>
</HTML>
------_=_NextPart_001_01C1679D.C098CB90--