Re: [Gnumed-devel] Corrupt database

[Busser, Jim]

On 2013-05-31, at 2:03 AM, Karsten Hilbert <address@hidden> wrote:

> On Fri, May 31, 2013 at 10:52:58AM +0200, Karsten Hilbert wrote:
> 
>> 3) we cannot simply use pg_dumpall and throw away the result
>>   because we need not have sufficient priviledges
> 
> However, if our backup can complete successfully that means
> there does not seem to be corruption of data and indices at
> the file system and PG file structure levels (no use at the
> layout[=hash] or content[=nigh impossible] levels, though).

Therefore, a combination of

(1) the client connecting without complaint, and

(2) setting up the backup, and determining that it completes successfully and

(3) ideally, restoring from a backup onto a different machine and confirming 
that a client will run

provide a good measure, and

even (1) and (2) without 3 provide modest reassurance … ?

-- Jim