On Thu, Apr 21, 2011 at 06:46:25PM +0200, Julien Coloos wrote: > It may be possible that some other set of conditions would trigger a > similar issue. > We tried to think of a clean way to solve it, but provided the > complexity of that code we gave up (mainly fearing nasty side effetcs > and introducing more bugs).
Yeah, for sure. > Maybe that's the reason it was completely rewritten in cyrus 2.4.x > (which does not have that bug) ? ;) Correct. > Unfortunately, moving to 2.4.x version is not yet an option for us :( Why not? > So ... is there some cyrus guru expert out there with enough knowledge > of that part of the code to fix it in the 2.3.x version ? Yes - I already did once... and I'd rather not touch it with a bargepole again! Seriously, there's all sorts of issues with that bit of the code in 2.3.x, and this is just one of them. I'd be interested in knowing your reasons for not upgrading to 2.4, because I'd prefer to fix those than ever look at the 2.3 seen management code again! Bron.