Re: AW: startup of resource groups with one node down

Martin Bene wrote:

> > -----Ursprüngliche Nachricht-----
> > Von: Lars Marowsky-Bree [mailto:[email protected]], 06. Mai 2002 19:09
>
> > > My question is: do you have to bring up all nodes in the
> > > cluster just
> > > for FailSafe to continue operating normally?  A person
> > > maintaining a
> > > cluster will have to know this...
> >
> > Yes. You need to bring up the majority of the cluster; at
> > least that is my
> > understanding.
>
> For a two-node cluster, that's what the tie-breaking node in the ha-parameters should be for; if only one node is available, it must be the tie-breaking node for a valid membership to be formed. I'd expect having to reset the tie-breaker parameter to be the available node in case I try to start up with just one node.
>

FailSafe notion of tie-breaker is different. Tiebreaker node gets the first chance to
reset other node in a two node cluster in case of network partition.

>
> Sadly, things still don't work out quite as expected:
>
> * If I start out with both nodes available and shut one node down:
>
> Node Status is 1xUP, 1xDOWN; resource groups fail over to the available node; rescource groups CAN be taken offline and set online again without problems.
>
> * If I start with just one node, which is the tiebreaker node:

If only one node in the cluster is operational, start HA services only on that
node. When the other node is available, you can start HA services on that node. It
should rejoin the cluster.

Paddy

>
>
> Node Status is 1x UP, 1xDOWN; resource startup is attempted but fails. resource groups end up in ERROR state, can be taken offline but can NOT be set online manually. manual startup fails:
>
>         Node in failure domain is not in membership. FailSafe
>         daemon (ha_fsd) failed to online resource group (db).
>
> or , with a bit more info from the logfiles:
>
> Mon May  6 19:33:23.166 <N resgroupAdmin config 1964:0 ci_config_cdb.c:208> resgroupAdmin _RESOURCE_GROUP=db _RESOURCE_GROUP_ACTION=online _CLUSTER=webc
> Mon May  6 19:33:53.592 <N resgroupAdmin config 1964:0 resgroupAdmin.c:853> Node in failure domain is not in membership
> Mon May  6 19:33:53.618 <E resgroupAdmin config 1964:0 ci_config_cdb.c:232> CI_FAILURE, CLI private command: failed (FailSafe daemon (ha_fsd) failed to online resource group (db).)
>
> Somehow, I can't imagine that that's expected/desired behaviour :-)
>
> Bye, Martin
> _______________________________________________
> LinuxFailSafe mailing list
> [email protected]
> http://lists.community.tummy.com/mailman/listinfo/linuxfailsafe