[Pacemaker] Few questions
Dejan Muhamedagic
dejanmm at fastmail.fm
Mon Feb 9 10:21:25 UTC 2009
On Mon, Feb 09, 2009 at 03:09:40PM +0530, Romi Verma wrote:
> On Mon, Feb 9, 2009 at 2:56 PM, Dejan Muhamedagic <dejanmm at fastmail.fm>wrote:
>
> > On Mon, Feb 09, 2009 at 02:16:42PM +0530, Glory Smith wrote:
> > > On Mon, Feb 9, 2009 at 2:06 PM, Dejan Muhamedagic <dejanmm at fastmail.fm
> > >wrote:
> > >
> > > > Hi,
> > > >
> > > > On Mon, Feb 09, 2009 at 01:26:56PM +0530, Glory Smith wrote:
> > > > > Hi All,
> > > > > Anybody worked on stonith external/sbd?? if yes please reply to my
> > > > following
> > > > > mail.
> > > > >
> > > > > On Fri, Feb 6, 2009 at 2:27 PM, Glory Smith <xx2glory at gmail.com>
> > wrote:
> > > > >
> > > > > > Hi All,
> > > > > > I am New to this list. i have few questions related to stonith
> > > > external
> > > > > > sbd plugin .
> > > > > >
> > > > > > i saw that document says that read/write to thd sbd device is
> > single
> > > > point
> > > > > > of failure . if we configure multipath then this should not be SPOF
> > > > anymore
> > > > > > . Right??
> > > >
> > > > Right. Provided your local rat community doesn't consider the
> > > > cables edible.
> > > >
> > > > > > it says that other nodes can write fence request to the errant
> > node's
> > > > > > mailbox and the errant node gets fenced.
> > > > > > can any one explain when it will happen .
> > > >
> > > > Whenever the CRM decides a node should be fenced. It has nothing
> > > > to do with sbd.
> > > >
> > > > > > According to my understanding so far, as long as the node is having
> > > > access
> > > > > > to sbd device it will not get fenced and once it looses access to
> > the
> > > > sbd
> > > > > > device it will reset itself . so when other nodes will get chance
> > to
> > > > fence
> > > > > > a node by writing fence request?
> > > >
> > > > Not true. A node may also get a "poison pill" from another node,
> > > > just as you mentioned yourself in the previous paragraph.
> > >
> > >
> > > As you said if CRM decides a node should be fenced it will be fenced and
> > it
> > > has nothing to do with sbd.
> >
> > Right. The decision itself has nothing to do with sbd. sbd is
> > just an executioner.
> >
> > > so in this case cluster must need an another
> > > stonith to provide fencing of errant node. is my understanding is
> > correct.
> >
> > No.
> >
> > > can you explain a bit more , how CRM decide that a node should be fence ?
> > is
> > > it when node become unreachable.??
> >
> > Yes. Also if a resource can't be stopped on that node.
>
>
>
> sorry to interrupt in between. just one question from my side. if
> no-quorum-policy is set to freeze then resource is not supposed to stop .
right.
> in
> that case node will not be fenced right??
no-quorum-policy has nothing to do with fencing.
Thanks,
Dejan
More information about the Pacemaker
mailing list