Joao Pereira wrote:
Shouldn't multimaster work??????
it is not guaranteed
ive compilled with --enable-multimaster.....
good it still compiles...
and would need it working like that
in any case, you cannot use the updatedn to directly write to a
multimaster. writes by the updatedn must come from the replication
agent, and they are supposed to contain the structuralObjectClass
along with other non-user modifiable data. If you write with that
identity, writes do not propagate to other masters in the pool.
Otherwise, how do you suppose a multimaster pool doesn't get into an
endless replication loop?
p.
SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497