On 12/20/2012 11:46 AM, Russ Allbery wrote:
>
> He has an existing cell with supergroups enabled and a bunch of
> supergroups in use.  He accidentally upgraded to a ptserver without
> supergroup support.  He's since replaced that with one that does have
> supergroup support, but he's wondering if the ptserver without supergroup
> support could have corrupted the supergroup portions of his database while
> it was running.
> 
> I wasn't sure -- I'd heard in the past that people were worried about what
> the ptserver might do to the database, but I don't know that specifics
> were ever discussed.

Jeff Hutzelman and I discussed this situation earlier this week.  The
conclusion was that deletions of groups with groups as members or as
members of a group can produce dangling references which:

 a. can generate inconsistent results

 b. may confuse tools that manage memberships

If there were no deletions while the supergroup unaware ptserver was
master, there should be no problems.  If there were (or may have been)
deletions, someone should write a tool to restore consistency to
the supergroup relationships and recover unreferenced but used blocks.

Jeffrey Altman


Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to