[fhs-discuss] HA failover hierarchy?

[Peter]

There are many tools, such as linux-ha, redhat-piranha, that do high
availability and failover.  The normal FHS calls for application data to be
in /var, but having that move onto the failover partner is, imo, a bad idea.
 I don't want /var/logs, for example not being available to the ha partner
which isn't running any services at the moment.

Currently, the FHS compliant method to deal with this would be to have /var
have a whole lot of symbolic links to an undefined location (some have an
internal standard of /apps) where the apps that failover store their data.
 It seems like a lot of sys admins have to make a lot of configurations.  It
can stay like that, but I wonder...

Should the new FHS have a new hierachy for application data that will
failover to the other side.  Should application data be moved out of /var to
enable transparent use of some other directory, like /apps, or /data. /var
would be reserved for explicitly node local activities (every node in an HA
cluster has it's own /var), whereas /apps would be for apps which are
expected to switchover, and have /apps appear on the other node.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.linux-foundation.org/pipermail/fhs-discuss/attachments/20110507/c96191d5/attachment.htm