"Dave Page" <dpage at vale-housing.co.uk> writes:
>> [mailto:slony1-general-bounces at gborg.postgresql.org] On 
>> Behalf Of Niels Breet
>> 
>> The problem with pgFoundry atm is this: (correct me if I'm 
>> wrong, this all
>> what I gathered from IRC chats)
>> - current jail runs on a server that is not controlled by the 
>> community?
>
> Incorrect. 
>
>> - current server is overloaded/ not reliable
>
> Incorrect. It was overloaded, but has long since been moved to a new
> box. There is a new dedicated box waiting in the wings though for
> someone with enough gforge-fu (and the time) to do the move.
>
>> - nobody has access to the database it is running on.
>
> Incorrect.
>
>> - it is running outdated and abandonned software. (Although 
>> that can be
>>   fixed by installing the non-open version?)
>
> It is running an old release. The move to the new server was/is supposed
> to include an upgrade to the latest FOSS release I believe.
>
>> - the replacement server has hardware problems that haven't 
>> been fixed for
>>   a while.
>
> Incorrect as far as I'm aware.

If Niels is wrong about all of these things, then there's a rather
serious communications problem, because I and others were also hearing
the same reports.

Are we merely "wrong" because these problems have been (silently)
repaired over the last few days?  If that's the case, then, in truth,
we *haven't* been misinformed.  If these things *were* true, and there
was no announcement of a fix, then for people to believe that these
conditions might continue to persist is in no way unreasonable.

Alternatively, are we "wrong" because these things *never* were true?
If that's the case, then someone spread outright untruth, which is a
different kind of distressing.

>> - nobody has emergency access to the upstream provider?
>
> That has been addressed.
>
>> What I gathered from the discussions on the needs of Slony-I is this:
>> - Remote backup (dataloss is not acceptable)
>
> Both Gborg and pgFoundry are currently backed up to another onsite
> server as well as an off-site. The new pgFoundry box is also being setup
> to hold additional backups.

Focusing a bit on this one point, what's wanted is the ability for
community members to draw backups to eliminate this as a single point
of failure.  If several organizations can pull backups, whereever the
server may be, that can adequately answer that issue.

The "single point of failure" isssue is one that has been applying to
far too many things:
 - Only one person with root access to systems
 - Only one person with control of DNS
 - Only one person with access to the backups

I'm not hearing things that convince me that there's not still a
single point of failure.  If others generally *do* feel convinced,
then perhaps it is reasonable for the migration to be to pgFoundry.

Non-negotiable, even in that scope, is that I want the ability to pull
backups of the CVS repository frequently (rsync can make that
reasonably efficient), as my level of trust in the systems hosted at
Hub has been seriously degraded by the events of the last month.

Simply saying that things have been addressed *isn't* enough, after
having been burnt by problems.
-- 
(format nil "~S@~S" "cbbrowne" "ca.afilias.info")
<http://dba2.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)