Thread: Time to get infrastructure team-based

Time to get infrastructure team-based

From
Bruce Momjian
Date:
I think we have seen enough problems with our email infrastructure that
we need to get a team that can do _everything_ related to email.

Right now Marc does it all and I think almost everyone agrees that
should change.  How do we move forward with this?

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
"Dave Page"
Date:
On Thu, Mar 20, 2008 at 9:47 PM, Bruce Momjian <bruce@momjian.us> wrote:
> I think we have seen enough problems with our email infrastructure that
>  we need to get a team that can do _everything_ related to email.
>
>  Right now Marc does it all and I think almost everyone agrees that
>  should change.  How do we move forward with this?

I've put forward suggestions on that topic in -core on numerous
occasions recently. Maybe you could comment on one of those threads?

-- 
Dave Page
EnterpriseDB UK Ltd: http://www.enterprisedb.com
PostgreSQL UK 2008 Conference: http://www.postgresql.org.uk


Re: Time to get infrastructure team-based

From
Bruce Momjian
Date:
Dave Page wrote:
> On Thu, Mar 20, 2008 at 9:47 PM, Bruce Momjian <bruce@momjian.us> wrote:
> > I think we have seen enough problems with our email infrastructure that
> >  we need to get a team that can do _everything_ related to email.
> >
> >  Right now Marc does it all and I think almost everyone agrees that
> >  should change.  How do we move forward with this?
> 
> I've put forward suggestions on that topic in -core on numerous
> occasions recently. Maybe you could comment on one of those threads?

I saw your request for documentation on the infrastructure but I don't
remember anything else.  I don't se this as a core discussion because
the team is going to be more than core folks.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Thu, 20 Mar 2008 21:55:26 +0000
"Dave Page" <dpage@pgadmin.org> wrote:

> On Thu, Mar 20, 2008 at 9:47 PM, Bruce Momjian <bruce@momjian.us>
> wrote:
> > I think we have seen enough problems with our email infrastructure
> > that we need to get a team that can do _everything_ related to
> > email.
> >
> >  Right now Marc does it all and I think almost everyone agrees that
> >  should change.  How do we move forward with this?
> 
> I've put forward suggestions on that topic in -core on numerous
> occasions recently. Maybe you could comment on one of those threads?
> 

Isn't that the point of sysadmins?

Sincerely,

Joshua D. Drake

- -- 
The PostgreSQL Company since 1997: http://www.commandprompt.com/ 
PostgreSQL Community Conference: http://www.postgresqlconference.org/
United States PostgreSQL Association: http://www.postgresql.us/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFH4t/aATb/zqfZUUQRAqyNAJwOmaOphEkIs6LFJaNvBR2QXsj/xACcCn5E
KO89HvbuOetk+xHCxFia8z0=
=556z
-----END PGP SIGNATURE-----

Re: Time to get infrastructure team-based

From
"Dave Page"
Date:
On Thu, Mar 20, 2008 at 10:00 PM, Bruce Momjian <bruce@momjian.us> wrote:
>  I saw your request for documentation on the infrastructure but I don't
>  remember anything else.  I don't se this as a core discussion because
>  the team is going to be more than core folks.

I've suggested a number of times that we need to decouple all
postgresql.org mail infrastructure from hub.org so it can be
independently managed. Numerous people have been asked the same thing
here for much longer, but have given up because the answer has always
been no. Thats why I've been raising it in core.

-- 
Dave Page
EnterpriseDB UK Ltd: http://www.enterprisedb.com
PostgreSQL UK 2008 Conference: http://www.postgresql.org.uk


Re: Time to get infrastructure team-based

From
Bruce Momjian
Date:
Dave Page wrote:
> On Thu, Mar 20, 2008 at 10:00 PM, Bruce Momjian <bruce@momjian.us> wrote:
> >  I saw your request for documentation on the infrastructure but I don't
> >  remember anything else.  I don't se this as a core discussion because
> >  the team is going to be more than core folks.
> 
> I've suggested a number of times that we need to decouple all
> postgresql.org mail infrastructure from hub.org so it can be
> independently managed. Numerous people have been asked the same thing
> here for much longer, but have given up because the answer has always
> been no. Thats why I've been raising it in core.

OK, so maybe there is the crux.  Who says "no" unchallenged?  I don't
think I can, and I don't think anyone else in this community can either.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
"Dave Page"
Date:
On Thu, Mar 20, 2008 at 10:22 PM, Bruce Momjian <bruce@momjian.us> wrote:
> Dave Page wrote:
>  > On Thu, Mar 20, 2008 at 10:00 PM, Bruce Momjian <bruce@momjian.us> wrote:
>  > >  I saw your request for documentation on the infrastructure but I don't
>  > >  remember anything else.  I don't se this as a core discussion because
>  > >  the team is going to be more than core folks.
>  >
>  > I've suggested a number of times that we need to decouple all
>  > postgresql.org mail infrastructure from hub.org so it can be
>  > independently managed. Numerous people have been asked the same thing
>  > here for much longer, but have given up because the answer has always
>  > been no. Thats why I've been raising it in core.
>
>  OK, so maybe there is the crux.  Who says "no" unchallenged?  I don't
>  think I can, and I don't think anyone else in this community can either.

I think you can guess who says no (if not, I'll tell you out of band).
I seem to be the only person doing any challenging and without some
nods of agreement at least I cannot do much more. We do at least have
some more documentation as of yesterday though... but the current
problems are with a whole server which that doesn't help with.

-- 
Dave Page
EnterpriseDB UK Ltd: http://www.enterprisedb.com
PostgreSQL UK 2008 Conference: http://www.postgresql.org.uk


Re: Time to get infrastructure team-based

From
Robert Treat
Date:
On Thursday 20 March 2008 18:26, Dave Page wrote:
> On Thu, Mar 20, 2008 at 10:22 PM, Bruce Momjian <bruce@momjian.us> wrote:
> > Dave Page wrote:
> >  > On Thu, Mar 20, 2008 at 10:00 PM, Bruce Momjian <bruce@momjian.us> 
wrote:
> >  > >  I saw your request for documentation on the infrastructure but I
> >  > > don't remember anything else.  I don't se this as a core discussion
> >  > > because the team is going to be more than core folks.
> >  >
> >  > I've suggested a number of times that we need to decouple all
> >  > postgresql.org mail infrastructure from hub.org so it can be
> >  > independently managed. Numerous people have been asked the same thing
> >  > here for much longer, but have given up because the answer has always
> >  > been no. Thats why I've been raising it in core.
> >
> >  OK, so maybe there is the crux.  Who says "no" unchallenged?  I don't
> >  think I can, and I don't think anyone else in this community can either.
>
> I think you can guess who says no (if not, I'll tell you out of band).
> I seem to be the only person doing any challenging and without some
> nods of agreement at least I cannot do much more. We do at least have
> some more documentation as of yesterday though... but the current
> problems are with a whole server which that doesn't help with.
>

If we're serious about doing this decoupling, I'll mention again that OmniTI 
would be interested in helping out... fwiw we helped out the PHP project 
setup thier current infrastructure.... and have a fair amount of knowledge in 
the email department. 

-- 
Robert Treat
Build A Brighter LAMP :: Linux Apache {middleware} PostgreSQL


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Thu, 20 Mar 2008 23:15:57 -0400
Robert Treat <xzilla@users.sourceforge.net> wrote:

> > I think you can guess who says no (if not, I'll tell you out of
> > band). I seem to be the only person doing any challenging and
> > without some nods of agreement at least I cannot do much more. We
> > do at least have some more documentation as of yesterday though...
> > but the current problems are with a whole server which that doesn't
> > help with.
> >
> 
> If we're serious about doing this decoupling, I'll mention again that
> OmniTI would be interested in helping out... fwiw we helped out the
> PHP project setup thier current infrastructure.... and have a fair
> amount of knowledge in the email department. 
> 

As a note to this :) I have a customer who swears buy CMD for
postgresql and OMNITI for their email products. I would have zero
reservation allowing OmniTI to host our mail infrastructure (assuming
it could be community managed).

Sincerely,

Joshua D. Drake


- -- 
The PostgreSQL Company since 1997: http://www.commandprompt.com/ 
PostgreSQL Community Conference: http://www.postgresqlconference.org/
United States PostgreSQL Association: http://www.postgresql.us/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFH5FRMATb/zqfZUUQRAubWAJ9E9JkuVy4N0/QbvCWz7gQIHGVt/QCfYf7E
GR8qBoAst67ZR7HrytgsO6M=
=6NKf
-----END PGP SIGNATURE-----

Re: Time to get infrastructure team-based

From
Bruce Momjian
Date:
Joshua D. Drake wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> On Thu, 20 Mar 2008 23:15:57 -0400
> Robert Treat <xzilla@users.sourceforge.net> wrote:
> 
> > > I think you can guess who says no (if not, I'll tell you out of
> > > band). I seem to be the only person doing any challenging and
> > > without some nods of agreement at least I cannot do much more. We
> > > do at least have some more documentation as of yesterday though...
> > > but the current problems are with a whole server which that doesn't
> > > help with.
> > >
> > 
> > If we're serious about doing this decoupling, I'll mention again that
> > OmniTI would be interested in helping out... fwiw we helped out the
> > PHP project setup thier current infrastructure.... and have a fair
> > amount of knowledge in the email department. 
> > 
> 
> As a note to this :) I have a customer who swears buy CMD for
> postgresql and OMNITI for their email products. I would have zero
> reservation allowing OmniTI to host our mail infrastructure (assuming
> it could be community managed).

Everyone here seems to be in agreement.  So what are the next steps? 
Setup a parallel email infrastructure, test it, and switch over?  What
else needs to be moved?

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
"Dave Page"
Date:
On Sat, Mar 22, 2008 at 2:07 PM, Bruce Momjian <bruce@momjian.us> wrote:

>  Everyone here seems to be in agreement.  So what are the next steps?
>  Setup a parallel email infrastructure, test it, and switch over?  What
>  else needs to be moved?

It's not exactly that easy because there is a complex mix of email
accounts, majordomo lists, anti-spam and anti-virus, and integration
with the archives system to take into account.

Once we have agreement from the appropriate person (who will have to
agree, given that he owns the domain), the sysadmins team can start to
figure out what's needed and provision the appropriate VM.

Note that any new mail infrastructure *will* be managed by the
sysadmin team as part of the existing FreeBSD infrastructure. Doing it
any other way invalidates most of the arguments for moving it in the
first place.


-- 
Dave Page
EnterpriseDB UK Ltd: http://www.enterprisedb.com
PostgreSQL UK 2008 Conference: http://www.postgresql.org.uk


Re: Time to get infrastructure team-based

From
Bruce Momjian
Date:
Dave Page wrote:
> On Sat, Mar 22, 2008 at 2:07 PM, Bruce Momjian <bruce@momjian.us> wrote:
> 
> >  Everyone here seems to be in agreement.  So what are the next steps?
> >  Setup a parallel email infrastructure, test it, and switch over?  What
> >  else needs to be moved?
> 
> It's not exactly that easy because there is a complex mix of email
> accounts, majordomo lists, anti-spam and anti-virus, and integration
> with the archives system to take into account.
> 
> Once we have agreement from the appropriate person (who will have to
> agree, given that he owns the domain), the sysadmins team can start to
> figure out what's needed and provision the appropriate VM.

Based on past history I don't think we will ever get agreement, so
either we do it ourselves or we move to a new domain that we do control.

I thought we could make changes to the domain.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
"Dave Page"
Date:
On Sat, Mar 22, 2008 at 2:21 PM, Bruce Momjian <bruce@momjian.us> wrote:
>  Based on past history I don't think we will ever get agreement, so
>  either we do it ourselves or we move to a new domain that we do control.
>
>  I thought we could make changes to the domain.

I can, but it can be changed so that I cannot in seconds. We *need*
agreement otherwise this could get extremely messy. Has anyone from
-core other than me actually asked yet?


-- 
Dave Page
EnterpriseDB UK Ltd: http://www.enterprisedb.com
PostgreSQL UK 2008 Conference: http://www.postgresql.org.uk


Re: Time to get infrastructure team-based

From
Bruce Momjian
Date:
Dave Page wrote:
> On Sat, Mar 22, 2008 at 2:21 PM, Bruce Momjian <bruce@momjian.us> wrote:
> >  Based on past history I don't think we will ever get agreement, so
> >  either we do it ourselves or we move to a new domain that we do control.
> >
> >  I thought we could make changes to the domain.
> 
> I can, but it can be changed so that I cannot in seconds. We *need*
> agreement otherwise this could get extremely messy. Has anyone from
> -core other than me actually asked yet?

No, but it seems like a waste of time to even try.

One idea is to set up the duplicate infrastructure with a domain in
place, and if the domain change gets blocked, we just switch to the new
domain name.

I am ready for messy.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
"Dave Page"
Date:
On Sat, Mar 22, 2008 at 2:37 PM, Bruce Momjian <bruce@momjian.us> wrote:
>  > I can, but it can be changed so that I cannot in seconds. We *need*
>  > agreement otherwise this could get extremely messy. Has anyone from
>  > -core other than me actually asked yet?
>
>  No, but it seems like a waste of time to even try.
>
>  One idea is to set up the duplicate infrastructure with a domain in
>  place, and if the domain change gets blocked, we just switch to the new
>  domain name.

I don't even want to think about the ramifications of doing that.


-- 
Dave Page
EnterpriseDB UK Ltd: http://www.enterprisedb.com
PostgreSQL UK 2008 Conference: http://www.postgresql.org.uk


Re: Time to get infrastructure team-based

From
Bruce Momjian
Date:
Dave Page wrote:
> On Sat, Mar 22, 2008 at 2:37 PM, Bruce Momjian <bruce@momjian.us> wrote:
> >  > I can, but it can be changed so that I cannot in seconds. We *need*
> >  > agreement otherwise this could get extremely messy. Has anyone from
> >  > -core other than me actually asked yet?
> >
> >  No, but it seems like a waste of time to even try.
> >
> >  One idea is to set up the duplicate infrastructure with a domain in
> >  place, and if the domain change gets blocked, we just switch to the new
> >  domain name.
> 
> I don't even want to think about the ramifications of doing that.

Well, then we will forever be hostage to the will of the domain name
holder.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
 + If your life is a hard drive, Christ can be your backup. +


Re: Time to get infrastructure team-based

From
Magnus Hagander
Date:
Robert Treat wrote:
> On Thursday 20 March 2008 18:26, Dave Page wrote:
>> On Thu, Mar 20, 2008 at 10:22 PM, Bruce Momjian <bruce@momjian.us> wrote:
>>> Dave Page wrote:
>>>  > On Thu, Mar 20, 2008 at 10:00 PM, Bruce Momjian <bruce@momjian.us> 
> wrote:
>>>  > >  I saw your request for documentation on the infrastructure but I
>>>  > > don't remember anything else.  I don't se this as a core discussion
>>>  > > because the team is going to be more than core folks.
>>>  >
>>>  > I've suggested a number of times that we need to decouple all
>>>  > postgresql.org mail infrastructure from hub.org so it can be
>>>  > independently managed. Numerous people have been asked the same thing
>>>  > here for much longer, but have given up because the answer has always
>>>  > been no. Thats why I've been raising it in core.
>>>
>>>  OK, so maybe there is the crux.  Who says "no" unchallenged?  I don't
>>>  think I can, and I don't think anyone else in this community can either.
>> I think you can guess who says no (if not, I'll tell you out of band).
>> I seem to be the only person doing any challenging and without some
>> nods of agreement at least I cannot do much more. We do at least have
>> some more documentation as of yesterday though... but the current
>> problems are with a whole server which that doesn't help with.
>>
> 
> If we're serious about doing this decoupling, I'll mention again that OmniTI 
> would be interested in helping out... fwiw we helped out the PHP project 
> setup thier current infrastructure.... and have a fair amount of knowledge in 
> the email department. 


+42 on the decoupling (something I've been arguing for for years, so 
that can't come as a surprise). It will help us two-fold: First, it will 
make the system simpler, and thus easier to track errors in. Second, it 
will open it up for more people to be able to help out with fixing 
issues in it.

And when we do that, not using the expertise offered by OmniTI would be 
a big mistake. As long as they way it'd be done is for them to help us 
out with how to configure it within the parameters that our 
infrastructure works on now, and not just deliver a "dropin 
off-the-shelf solution" that we should be using. But with my 
understanding of how OmniTI works, I don't see that as being a problem.

//Magnus


Re: Time to get infrastructure team-based

From
Magnus Hagander
Date:
Bruce Momjian wrote:
> Dave Page wrote:
>> On Sat, Mar 22, 2008 at 2:21 PM, Bruce Momjian <bruce@momjian.us> wrote:
>>>  Based on past history I don't think we will ever get agreement, so
>>>  either we do it ourselves or we move to a new domain that we do control.
>>>
>>>  I thought we could make changes to the domain.
>> I can, but it can be changed so that I cannot in seconds. We *need*
>> agreement otherwise this could get extremely messy. Has anyone from
>> -core other than me actually asked yet?
> 
> No, but it seems like a waste of time to even try.

If -core can deliver an actual decision on this, I don't see why it 
wouldn't happen. Let's not assume failure before we've tried. But I do 
believe we need a decision from -core, just vocal members on this list 
is obviously not enough (because we've tried that).


//Magnus



Re: Time to get infrastructure team-based

From
Magnus Hagander
Date:
Bruce Momjian wrote:
> Dave Page wrote:
>> On Sat, Mar 22, 2008 at 2:37 PM, Bruce Momjian <bruce@momjian.us> wrote:
>>>  > I can, but it can be changed so that I cannot in seconds. We *need*
>>>  > agreement otherwise this could get extremely messy. Has anyone from
>>>  > -core other than me actually asked yet?
>>>
>>>  No, but it seems like a waste of time to even try.
>>>
>>>  One idea is to set up the duplicate infrastructure with a domain in
>>>  place, and if the domain change gets blocked, we just switch to the new
>>>  domain name.
>> I don't even want to think about the ramifications of doing that.
> 
> Well, then we will forever be hostage to the will of the domain name
> holder.

Yes, but that would be equally true of the holder of the new domain. It 
would, however, make sense to have the domain reassigned to 
postgresql.org - it's currently owned by pgsql,inc. But you still need a 
contacts on the domain that are individuals, who will always have at 
least a theoretical ability to "hold hostage".

But this is really a different question from the other one. I'm not 
saying both shouldn't be fixed, I'm just saying it's different things.

//Magnus


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
On Sat, 22 Mar 2008 14:14:36 +0000
"Dave Page" <dpage@pgadmin.org> wrote:

> Note that any new mail infrastructure *will* be managed by the
> sysadmin team as part of the existing FreeBSD infrastructure. Doing it
> any other way invalidates most of the arguments for moving it in the
> first place.

Right.

Joshua D. Drake


--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
On Sat, 22 Mar 2008 10:21:42 -0400 (EDT)
Bruce Momjian <bruce@momjian.us> wrote:
> > Once we have agreement from the appropriate person (who will have to
> > agree, given that he owns the domain), the sysadmins team can start
> > to figure out what's needed and provision the appropriate VM.
>
> Based on past history I don't think we will ever get agreement, so
> either we do it ourselves or we move to a new domain that we do
> control.

I seriously doubt that given past history, the pressure required to
make such a change has ever been applied. If you are serious about
this, it must be a public change, it must be a documented change, and
the community as a whole needs to be involved.

This is either a "community" project or it isn't. People (any people)
that hold up the will of the community must be held accountable. Unless
the people actively suggesting this change are willing to start that
publicly accountable discussion, I suggest we drop this thread now.

>
> I thought we could make changes to the domain.
>

The domain is currently controlled by a single individual.

Sincerely,

Joshua D. Drake

--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
On Sat, 22 Mar 2008 10:37:01 -0400 (EDT)
Bruce Momjian <bruce@momjian.us> wrote:

> One idea is to set up the duplicate infrastructure with a domain in
> place, and if the domain change gets blocked, we just switch to the
> new domain name.
>
> I am ready for messy.
>

Bruce with respect, we obviously aren't because we aren't addressing
this problem on a list where the wider community understands the
issues. Heck we aren't even willing to lodge a professional complaint
(note word professional) that mentions the parties involved.

Joshua D. Drake

--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
On Sat, 22 Mar 2008 14:41:05 +0000
"Dave Page" <dpage@pgadmin.org> wrote:

> >  No, but it seems like a waste of time to even try.
> >
> >  One idea is to set up the duplicate infrastructure with a domain in
> >  place, and if the domain change gets blocked, we just switch to
> > the new domain name.
>
> I don't even want to think about the ramifications of doing that.

It would be ugly.

Joshua D. Drake

--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit


Re: Time to get infrastructure team-based

From
"Joshua D. Drake"
Date:
On Sat, 22 Mar 2008 16:34:38 +0100
Magnus Hagander <magnus@hagander.net> wrote:

> > Well, then we will forever be hostage to the will of the domain name
> > holder.
>
> Yes, but that would be equally true of the holder of the new domain.
> It would, however, make sense to have the domain reassigned to
> postgresql.org - it's currently owned by pgsql,inc. But you still

We can create a trustee agreement for this sort of thing.

Joshua D. Drake



--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL SPI Liaison | SPI Director |  PostgreSQL political pundit


Re: Time to get infrastructure team-based

From
Andreas 'ads' Scherbaum
Date:
Hello all,

On Thu, 20 Mar 2008 22:07:28 +0000 Dave Page wrote:

> On Thu, Mar 20, 2008 at 10:00 PM, Bruce Momjian <bruce@momjian.us> wrote:
> >  I saw your request for documentation on the infrastructure but I don't
> >  remember anything else.  I don't se this as a core discussion because
> >  the team is going to be more than core folks.
> 
> I've suggested a number of times that we need to decouple all
> postgresql.org mail infrastructure from hub.org so it can be
> independently managed. Numerous people have been asked the same thing
> here for much longer, but have given up because the answer has always
> been no. Thats why I've been raising it in core.

It doesn't have to be decoupled - but given the problems (plural) in
the past, it really makes sense. I remember uncounted times with delayed
mails, not just minutes but hours. I even sometimes got answers on my
mails hours before i have seen my own mails on the list. The delay was
always, according to the Received headers, a hub.org mail server.

Oh, and for
http://archives.postgresql.org/pgsql-www/2008-02/msg00219.php
i'm still waiting for an answer. I just got word, that i was not the
only one with this problem.

I don't know, how many problems it needs to drive away people from
PostgreSQL, but if we start to unsubscribe people from the mailinglists,
most of them won't come back.

So my personal opinion is: the PostgreSQL project should have his own
infractructure for mail and mailinglists, because in this case the
sysadmin team can take care in case of a problem.

This are just my 0.02 Euro-cent, i'm neither -core nor -www, i'm
just *censored* about the current situation.


Kind regards

--             Andreas 'ads' Scherbaum
German PostgreSQL User Group
European PostgreSQL User Group - Board of Directors


unsubscribes (wasRe: Time to get infrastructure team-based)

From
"Marc G. Fournier"
Date:
> Oh, and for 
> http://archives.postgresql.org/pgsql-www/2008-02/msg00219.php i'm still 
> waiting for an answer. I just got word, that i was not the only one with 
> this problem.

Sorry, thought i had responded to this one ... It's already well discussed 
on the -core list, but, around the 25th, I went through some of the 
various logs and built up a list of 'no such user' and 'delivery refused' 
addresses and unregistered them, to clean off the lists of dead accounts 
... although I scanned through the list before doing it, I should have 
gotten a second set of eyes to look it over ... manual failure on my part, 
sorry :(



Re: unsubscribes (wasRe: Time to get infrastructure team-based)

From
Andreas 'ads' Scherbaum
Date:
Hello,

On Sat, 22 Mar 2008 17:32:49 -0300 (ADT) Marc G. Fournier wrote:

> > Oh, and for 
> > http://archives.postgresql.org/pgsql-www/2008-02/msg00219.php i'm still 
> > waiting for an answer. I just got word, that i was not the only one with 
> > this problem.
> 
> Sorry, thought i had responded to this one ... It's already well discussed 
> on the -core list

I don't read -core, as most of the other concerned don't do.
And it was just one example which happened to me, more often i see other
people writing about mailing list problems on IRC.


> built up a list of 'no such user' and 'delivery refused' addresses

I remember that the hub.org DNS had some problems in the past (invalid
or not working reverse, if i remember correctly) which caused rejects
on my mail server for mail from hub.org mail server.
Just don't wanted to add this one too in my previous mail.


> to clean off the lists of dead accounts 

A good mailing list software can take care about this problem for you, i
don't know, if majordomo has such an option. If not, that feature would
be on my wishlist for next christmas.


> sorry :(

Accepted. As long as i know the reason, it's ok.


Kind regards

--             Andreas 'ads' Scherbaum
German PostgreSQL User Group
European PostgreSQL User Group - Board of Directors


Re: unsubscribes (wasRe: Time to get infrastructure team-based)

From
"Marc G. Fournier"
Date:
On Sun, 23 Mar 2008, Andreas 'ads' Scherbaum wrote:

>
> Hello,
>
> On Sat, 22 Mar 2008 17:32:49 -0300 (ADT) Marc G. Fournier wrote:
>
>>> Oh, and for
>>> http://archives.postgresql.org/pgsql-www/2008-02/msg00219.php i'm still
>>> waiting for an answer. I just got word, that i was not the only one with
>>> this problem.
>>
>> Sorry, thought i had responded to this one ... It's already well discussed
>> on the -core list
>
> I don't read -core, as most of the other concerned don't do.

Actually, I thought I had posted it here ... apparently I didn't ...

> I remember that the hub.org DNS had some problems in the past (invalid
> or not working reverse, if i remember correctly) which caused rejects
> on my mail server for mail from hub.org mail server.

Actually, the only issue in that past that caused us this sort of issue 
was using the delivery relays, where outgoing email was relayed through a 
server whose reverse dns, as you point out, didn't match the forward ...

... since we are going to be expanding the delivery system to allow for 
more regional relays, this is something that I have added to our checklist 
on the new setup, thank you for the reminder of that one...


Re: Time to get infrastructure team-based

From
Robert Treat
Date:
On Saturday 22 March 2008 12:13, Andreas 'ads' Scherbaum wrote:
> Oh, and for
> http://archives.postgresql.org/pgsql-www/2008-02/msg00219.php
> i'm still waiting for an answer. I just got word, that i was not the
> only one with this problem.
>

Hmm... seems I completly missed that email since I was going through the same 
issue, otherwise I would have piled on... almost funny if it weren't so #@!%^ 
annoying... 

-- 
Robert Treat
Build A Brighter LAMP :: Linux Apache {middleware} PostgreSQL


Re: unsubscribes (wasRe: Time to get infrastructure team-based)

From
Alvaro Herrera
Date:
Andreas 'ads' Scherbaum wrote:

> > to clean off the lists of dead accounts 
> 
> A good mailing list software can take care about this problem for you, i
> don't know, if majordomo has such an option. If not, that feature would
> be on my wishlist for next christmas.

Majordomo certainly has this feature.  However, a list owner must still
keep an eye on bounce reports because sometimes users have set up
forwardings or redistribute list posts in unusual ways, which then
generate bounces that the bounce processor cannot handle.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support


Re: unsubscribes (wasRe: Time to get infrastructure team-based)

From
Andrew Sullivan
Date:
On Sat, Mar 22, 2008 at 08:54:57PM -0300, Marc G. Fournier wrote:

> >or not working reverse, if i remember correctly) which caused rejects
> >on my mail server for mail from hub.org mail server.
> 
> Actually, the only issue in that past that caused us this sort of issue 
> was using the delivery relays, where outgoing email was relayed through a 
> server whose reverse dns, as you point out, didn't match the forward ...

Please note that there is an Internet Draft currently in Working Group Last
Call that suggests among other things that rejecting email because the
reverse doesn't work is quite possibly not ideal.

http://tools.ietf.org/wg/dnsop/draft-ietf-dnsop-reverse-mapping-considerations/

Comment on the dnsop working group mailing list would be welcome.

A