[Date Prev][Date Next]
Re: Robustness of replication
Quanah Gibson-Mount wrote:
>>> delta-syncrepl has been very solid for me, with the exception of the
>>> occasional edge case. The last time there was an edge case encountered
>>> was with 2.4.22 (4/24/2010).
>> Are you running with slapo-memberof?
>> (see also ITS#6915 since delta-syncrepl needs slapo-accesslog)
> No, I'm not. But as noted in the ITS, that's a long standing bug in
> slapo-memberof, and not in delta-syncrepl. It has in fact existed since
> slapo-memberof was first written. ;)
I've also read the ITS carefully. And I've observed my logdb!
The point is that the operations by slapo-memberof reach the logdb but the
write operation to the group entry is not in the logdb! So from my
understanding delta-syncrepl won't work correctly if slapo-memberof is used on
the provider no matter in which part of the code the bug was.
So the general recommendation for delta-syncrepl has some caveats...