[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [Linux-cluster] Two-node cluster: Node attempts stateful merge after clean reboot



On Thu, Sep 12, 2013 at 12:21:28AM +0200, emmanuel segura wrote:
> clean_start=1 disable the startup fencing and if you use a quorum disk in your
> cluster without expected_votes=1, when a node start after it has been fenced,
> the node dosn't try to fence di remain node and try to start the service,
> because rgmanager need a cluster quorate, so many people around say clean_start
> =1 is dangerous, but no one give a clear reason,

Listen to Digimer.  clean_start=1 is dangerous.  It will allow a node
to join a cluster "with state" and thus opens the door to split-brain.
We (Red Hat) will not support a production cluster with clean_start
turned on.

>                                                  in my production cluster a i
> have clvm+vg in exclusive mode+(clean_start=1)+(master_wins). so if you can
> explain me where is the problem :) i apriciate
> 

It specifically targets a safety mechanism, and basically turns
off the check for a node trying to join the cluster "with state",
so it will gladly allow you to split-brain a FS or something really
ugly like that.  It's there for testing/debugging purposes only,
and should never be used on a production system.

As for leaving services turned off, Digimer is spot on with that one,
too (that was you, wasn't it?).  It is one of the ways we recommend
getting around this fence loop problem.  It's also the simplest one
and, IMHO, the one with the most tolerable list of potential side
effects, which again Digimer did a fine job of explaining (basically
you have to start your cluster services manually, but if you have a
fence event, you're going to probably be fixing something anyway on
that machine, so, it's probably good that they're not coming up on
their own, and it's no big thing to start things up when you're done).

L8r,
tg.

> 2013/9/11 Digimer <lists alteeve ca>
> 
>     On 11/09/13 12:04, emmanuel segura wrote:
> 
>         Hello Pascal
> 
>         For disable startup fencing you need clean_start=1 in the fence_daemon
>         tag, i saw in your previous mail you are using expected_votes="1", with
>         this setting every cluster node will be partitioned into two clusters
>         and operate independently, i recommended using a quorim disk with
>         master_wins parameter
> 
> 
>     This is a very bad idea and is asking for a split-brain, the main reason
>     fencing exists at all.
> 
> 
>     --
>     Digimer
>     Papers and Projects: https://alteeve.ca/w/
>     What if the cure for cancer is trapped in the mind of a person without
>     access to education?
> 
> 
> 
> 
> --
> esta es mi vida e me la vivo hasta que dios quiera

> -- 
> Linux-cluster mailing list
> Linux-cluster redhat com
> https://www.redhat.com/mailman/listinfo/linux-cluster


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]