AW: AW: more corruption - Mailing list pgsql-hackers

From Zeugswetter Andreas SB
Subject AW: AW: more corruption
Date
Msg-id 11C1E6749A55D411A9670001FA687963367FEC@sdexcsrv1.f000.d0188.sd.spardat.at
Whole thread Raw
List pgsql-hackers
> > > > > -currently vacuuming the new db - nothing is barfing yet
> > 
> > Actually, the vacuum seemed to be running forever making no 
> > progress so
> > I killed it.
> > 
> > > > > -now hopefully I can create my indexes and be back in business
> > 
> > I vacuumed here and it worked. I did not use my "old" pg_log 
> > file - what
> > did I lose?
> 
> Tuples that have been inserted/updated/deleted after last vacuum
> in old db assume that the corresponding txn has to be rolled back.

Correction: Tuples that have been inserted/updated/deleted but have not
been accessed afterwards (the magic first access that updates the tuple 
transaction status inplace).

> Since your vacuum on old db only did half the db, that half 
> will be current,
> but the rest will be old, thus you loose consistency.
> 
> One of the core please confirm.

Andreas


pgsql-hackers by date:

Previous
From: Karel Zak
Date:
Subject: memory: bug or feature
Next
From: darcy@druid.net (D'Arcy J.M. Cain)
Date:
Subject: Re: Re: postgres TODO