Re: [SSI-users] DRBD-Failover of non-root file systems

[Eduardo S. <edu...@gm...>]

Hi Ivan,

2005/12/3, Ivan Krstic <kr...@fa...>:
> Eduardo Spinola wrote:
> > but when we add node1 to cluster, the system not complete boot, trying
> > check filesystem(/dev/drbd/1), entering in bash to recover system, I
> > don't remember accurately, something about readonly filesystem.
>
> Are you sure that when you're booting node1, it's booting over the
> network, and not trying to boot from its hard drive, as it normally would=
?
>
Booting via network we don=B4t have problems, but without drbd activated.

> > Well, 2 each one for root failover nodes is sufficient :-). But we
> > have 5 servers(HP ProLiant DL380 that we will use in cluster) with 6
>
> I use the same servers with no problems.
>
We still testing on desktop PC.. :-)

> > hds each one And we want one way to use it and have HA and more
> > performance(for storage database files and more). How about Lustre
> > Filesystem, or openGFS? Is good?
>
> Are you certain a clustered filesystem is what you're looking for there?
>
>
DRBD wich 2 nodes for root failover is good for us.
But all these disks are more than we need.
Our bigger problem is with storage IO, and we want any way use these
HD's to have more performace and HA. To have more performance with
database(Mysql, postgre and Intersystems Cach=E9), webservices(apache,
tomcat), and more.

Sorry my my english(is very bad :-) ), and my little knowledge about
cluster and linux

thanks for all,

Eduardo Montes Spinola
Brazil