Re: [Sks-devel] unsynchronized (2005-01-11): pks.gpg.cz, sks.keyserver.penguin.de, gnupg.jccc.net

[Yaron Minsky]

On Wed, 12 Jan 2005 12:22:57 -0500, Jason Harris <address@hidden> wrote:
> On Wed, Jan 12, 2005 at 12:11:28PM +0100, Bjoern Buerger wrote:
> > Am Di, 11 Jan 2005 schrieb Jason Harris:
> 
> > > Looking at http://pgp.srv.ualberta.ca/cgi-bin/sksnet , I see that
> > > pks.gpg.cz and sks.keyserver.penguin.de aren't currently synchronized.
> >
> > fixed. The PTree Database had been corrupted (probably by an unclean
> > shutdown of the sks process). I thought I had fixed this almost a week
> > before, but just running db_recover obviously didn't do the job.
> 
> Dbenv.RECOVER is set for the ./KDB/ and ./PTree/ database environments,
> so a regular recovery is always attempted by "sks db" and "sks recon."
> There is likely some other race condition that caused your error.
> 
> Yaron, are these messages:
> 
>   Added # hash-updates. Caught up to ...

That indicates the position of the PTree relative to a counter kept in
the Key database.  There are some situations where a PTree and the KDB
can get out of sync in a crash, so that the PTree is ahead of the KDB
after recovery.  In such cases, the right thing to do is to blow away
and rebuild the PTree.

I don't think the system clock should have any effect.

> indicitave of the actual PTree nodes and the timestamp being updated in
> the same transaction in the ptree db?  Also, what happens if the system
> clock is stepped backwards while the SKS processes are running?
> 
> --
> Jason Harris           |  NIC:  JH329, PGP:  This _is_ PGP-signed, isn't it?
> address@hidden _|_ web:  http://keyserver.kjsl.com/~jharris/
>           Got photons?   (TM), (C) 2004
> 
> 
> _______________________________________________
> Sks-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/sks-devel
> 
> 
> 
>