[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: ITS #2623 (out of order replog)



At 09:21 AM 11/12/2003, Frank Swasey wrote:
>I have just been bitten by this bug running 2.1.22 on RH9 where a modrdn
>was followed immediately by a modify to the master (which did both with
>no problems), but the replog was out of order and the modify was
>attempted on the slave before the modrdn was done.  Thus a reject file
>was created which when given to a oneshot slurpd worked.
>
>I see that ITS #2623 addresses this issue, but has been closed
>apparently saying that syncrepl is to be our salvation in this area.
>
>However, isn't syncrepl only on the 2.2 beta code?  Anyone have
>suggestions about how to minimize (or remove) the chances for this race
>to the replog in 2.1?

As noted in ITS#2623, the out-of-order replog problem is inherit
in design.  The fix is a redesigned replication system.

Kurt