drbd question

[Fábio Olivé Leite]

Hi there,

Philipp might be able to elaborate better on the subject, but here goes:

) Most of the discussion seemed to assume these two.  The only extension
) over what has been talked about all along for drbd was that it should
) somehow "detect and raise an error" for the multiple node failure
) situation.

Well, drbd needs to negotiate the shared disk size before it can operate,
so it either only operates _after_ a first connection has been
established, or if the sysadmin chooses to specify the disk size by hand,
in which case drbd can operate even if it has never seen the other end.

Since for the tests I have always set the disksize by hand, I forgot the
possibility that this might be used to protect shared data when both nodes
come up after a crash. If you don't specify disk size, it will only work
properly if there is always at least one node up. This node is obviuosly
going to have the TrueBits.

This is just an idea, I have to study the code better to be certain.

See ya
Fábio
( Fábio Olivé Leite -* ConectivaLinux *- olive@conectiva.com[.br] )
( Linux - Distributed Systems - Fault Tolerance - Security - /etc )