[DRBD-user] Recommendation on shared storage

Jose Perez jvoorhees1 at gmail.com
Tue Mar 17 22:35:00 CET 2009

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


hi people:

I pretend to build a HA cluster using Heartbeat with 2 nodes
(Active/Standby). These nodes are just compatible computers (PCs), so
they don't have any IPMI, RSA, APC or similar features to use stonith.

I'm thinking about using DRBD to replicate information between both
nodes so I would like someone with more experience than me give me
some advices to achieve a reliable setup.

- If I won't be able to use stonith to kill the other node in split
brain situation... Does my information will be at risk?
- If I use a dual-primary DRBD configuration... Is it possible that
data loss occurs in split brain situations? Do I need to use GFS/OCFS
in this scenario? How reliable is this for production use=
- What would be that appropiate shared storage configuration for my HA
Cluster? I have seen some posts talking about DRBD+NFS or DRBD+iSCSI..
How reliable are the alternatives that I have?

I'm trying to make the right choice of technologies for my cluster
scenario so I would appreciate some advices. Please consider my ONLY
TWO PCs available (both nodes), or maybe suggest me a better scenario
(maybe a third PC or something else) so I can evaluate the better
choice.

Thanks people , bye :)



More information about the drbd-user mailing list