reconstruct and uidvalidity

2003-10-28 Thread Kjetil Torgrim Homme
we recently had major problems with a disk system, and had to roll files back from tape. running "reconstruct -f -r user" mostly did the trick, but some users had issues with the clients not showing them new e-mail messages. what has happened, I think, is that they had messages 1,2,3 in their inb

Re: reconstruct and uidvalidity

2003-10-28 Thread Rob Siemborski
On Tue, 28 Oct 2003, Kjetil Torgrim Homme wrote: > assuming my understanding is correct, I think we need a flag to > reconstruct to force bumping up UIDVALIDITY. (a good thing Cyrus uses a > timestamp for this -- we don't have to worry about bumping it far > enough, it just will.) Removing the c

Re: reconstruct and uidvalidity

2003-10-28 Thread Kjetil Torgrim Homme
On Tue, 2003-10-28 at 16:45, Rob Siemborski wrote: > On Tue, 28 Oct 2003, Kjetil Torgrim Homme wrote: > > > assuming my understanding is correct, I think we need a flag to > > reconstruct to force bumping up UIDVALIDITY. (a good thing Cyrus uses a > > timestamp for this -- we don't have to worry

Re: reconstruct and uidvalidity

2003-10-28 Thread Rob Siemborski
On Tue, 28 Oct 2003, Kjetil Torgrim Homme wrote: > > Removing the cyrus.index file should force this to happen. > > thank you for your response. > > I'm not clear on what the three files cyrus.{header,cache,index} > contain, but won't this lose the user's flags? (except \Seen) Yes, this will be