On Thu, 2007-09-06 at 12:21 +0200, apoc9009 wrote:
> If the System Crash, you have Dataloss of
> over 16 MBytes thats Fatal and not acceptable! 1MByte Dataloss of ASCII Data is also
> not acceptable!

Is any data loss acceptable in the case of a disaster? How much?

--
Simon Riggs
2ndQuadrant http://www.2ndQuadrant.com


---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

> So you want the user to still be connected to the failed machine, but at
> the same time be connected to the new live failover machine ?
>
> -----
> Hannu
No.
The User should be connected to the running db without restrictions
while backup is in progress

Apoc

---------------------------(end of broadcast)---------------------------
TIP 7: You can help support the PostgreSQL project by donating at

http://www.postgresql.org/about/donate

Ühel kenal päeval, N, 2007-09-06 kell 12:53, kirjutas apoc9009:
> I wish to have an Solution, thadt backup my Database DB wihout
> Datalosses,

Then you need the backup record to be completed on the backup machine
before the transaction commit returns on master.

This is quaranteed to be really slow for any significant transaction
rate but can be done using DRBD.

> without locking Tables,

No backup I know of needs locking tables

> without Shutdown

This one I just don't understand. Shtdown what ?

> without any User must be forced for logging out (Backup in
> Production State Online without Datalosses).

So you want the user to still be connected to the failed machine, but at
the same time be connected to the new live failover machine ?

-----
Hannu




---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

Ühel kenal päeval, N, 2007-09-06 kell 16:15, kirjutas apoc9009:
> > So you want the user to still be connected to the failed machine, but at
> > the same time be connected to the new live failover machine ?
> >
> > -----
> > Hannu
> No.
> The User should be connected to the running db without restrictions
> while backup is in progress

And what's wrong with WAL-based backup then ?

Especially the variant described in 23.4.4 wich keeps the WAL copied to
backup machine up to last second ?

---------------
Hannu



---------------------------(end of broadcast)---------------------------
TIP 7: You can help support the PostgreSQL project by donating at

http://www.postgresql.org/about/donate

On Thu, 2007-09-06 at 12:53 +0200, apoc9009 wrote:
> > You've either not read 23.4.4 or haven't understood it. If the text is
> > unclear, documentation additions/changes are always welcome.
> I have read this:
>
> PostgreSQL directly supports file-based log shipping as described above.
> It is also possible to implement record-based log shipping, though this
> requires custom development.

We may expect that feature in later releases, but definitely not in 8.3

> I wish to have an Solution, thadt backup my Database DB wihout
> Datalosses, without locking Tables, without Shutdown
> and without any User must be forced for logging out (Backup in
> Production State Online without Datalosses).

I'm not clear whether you are looking for Backup, or High Availability
Replication.

There is no data loss with the online backup technique described in the
manual.

If you require HA replication with zero data-loss then you are currently
restricted to non-database techniques, which you already mentioned, so
you have your answer already.

--
Simon Riggs
2ndQuadrant http://www.2ndQuadrant.com


---------------------------(end of broadcast)---------------------------
TIP 3: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faq

Heikki Linnakangas schrieb:
> apoc9009 wrote:
>
> "Without datalosses" is utopy. For that, you'd need something like
> synchronous replication, otherwise you're always going to have a window
> where you have something committed in the server, but not yet in the
> backup. So it's just a question of how wide that window is, how much
> data loss can you live with.
>
> With file-based log shipping, you can get down to 1 second, by using the
> archive_timeout setting. It will produce a lot of log files with very
> little content in them, but they will compress well.
>
Yes, it is possible but not recommended . My Backup Servers Filesystem
will explode

> The record-based log shipping will give you a very narrow window, down
> to < 1 second or even less if you're willing to poll the server that
> often, but requires some custom development.
>
Correct, but this is not good enought and i think there are a lot of
Peoples having the same Problem.
It was wishfull, having a Online Streaming Backupsubsystem, thadt can
produce portable Backupdatabases,
to prevent users for Millions of Archive Files on the Backup FTP-Server.

My Ideaa is the following Folder Structure:

/pg/backup/

Backup 12/24/2007 Version 1
/pg/backup/12_24_2007/base/rcvry.rcv <--- Basebackup
/pg/backup/12_24_2007/changes/0001.chg <--- Changed Data
/changes/0002.chg <--- Changed Data
/changes/0003.chg <--- Changed Data

Backup 12/24/2008 Version 2
/pg/backup/12_24_2008/base/rcvry.rcv <--- Basebackup
/pg/backup/12_24_2008/changes/0001.chg <--- Changed Data
/changes/0002.chg <--- Changed Data
/changes/0003.chg <--- Changed Data
....
/changes/0010.chg <--- Changed Data

/changes/0001.rsf <--- Recovery
Stripeset File (10 MByte) addon of Basebackup
delete *.chg

if a Stripeset of 10 *.chg Files exist, they should be converted or merged
to one greater Recovery Stripe File (*.RSF)

Things to do:

1.A Integradted FTP-Client to the Postmaster Process
2.Online Streamingbackup Logic inside the Postmaster

Apoc


---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo@postgresql.org so that your
message can get through to the mailing list cleanly

Simon Riggs schrieb:
> On Thu, 2007-09-06 at 12:53 +0200, apoc9009 wrote:
>
>>> You've either not read 23.4.4 or haven't understood it. If the text is
>>> unclear, documentation additions/changes are always welcome.
>>>
>> I have read this:
>>
>> PostgreSQL directly supports file-based log shipping as described above.
>> It is also possible to implement record-based log shipping, though this
>> requires custom development.
>>
>
> We may expect that feature in later releases, but definitely not in 8.3
>
>
>> I wish to have an Solution, thadt backup my Database DB wihout
>> Datalosses, without locking Tables, without Shutdown
>> and without any User must be forced for logging out (Backup in
>> Production State Online without Datalosses).
>>
>
> I'm not clear whether you are looking for Backup, or High Availability
> Replication.
>
> There is no data loss with the online backup technique described in the
> manual.
>
No, there is a lost of Data.

The WAL File musst bei generated by the Postmaster and this File must be
copied to the
safe Remote Backupserver. If the Machine, where the Database is running
crashed, then the Last
WAL is lost and the Backup isnt complet,e this is the Center of the Problem.
> If you require HA replication with zero data-loss then you are currently
> restricted to non-database techniques, which you already mentioned, so
> you have your answer already.
>
we talking about Backup this isnt the same as Replication.

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

On 9/6/07, apoc9009 <apoc9009@yahoo.de> wrote:

> Backup 12/24/2008 Version 2
> /pg/backup/12_24_2008/base/rcvry.rcv <--- Basebackup
> /pg/backup/12_24_2008/changes/0001.chg <--- Changed Data
> /changes/0002.chg <--- Changed Data
> /changes/0003.chg <--- Changed Data
> ....
> /changes/0010.chg <--- Changed Data
>
> /changes/0001.rsf <--- Recovery
> Stripeset File (10 MByte) addon of Basebackup
> delete *.chg
>
> if a Stripeset of 10 *.chg Files exist, they should be converted or merged
> to one greater Recovery Stripe File (*.RSF)

Why? What does this actually do?

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

On Thu, Sep 06, 2007 at 04:08:10PM +0200, apoc9009 wrote:
> >archive_timeout setting. It will produce a lot of log files with very
> >little content in them, but they will compress well.
> >
> Yes, it is possible but not recommended . My Backup Servers Filesystem
> will explode

.. . .

> Correct, but this is not good enought and i think there are a lot of
> Peoples having the same Problem.
> It was wishfull, having a Online Streaming Backupsubsystem, thadt can
> produce portable Backupdatabases,
> to prevent users for Millions of Archive Files on the Backup FTP-Server.

It seems that what you want is near-real-time online backups with _no
cost_, which is not a feature that I think anyone will ever work on.

A

--
Andrew Sullivan | ajs@crankycanuck.ca
Users never remark, "Wow, this software may be buggy and hard
to use, but at least there is a lot of code underneath."
--Damien Katz

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

On Thu, 2007-09-06 at 17:53 +0300, Hannu Krosing wrote:
> Ühel kenal päeval, N, 2007-09-06 kell 16:15, kirjutas apoc9009:
> > > So you want the user to still be connected to the failed machine, but at
> > > the same time be connected to the new live failover machine ?
> > >
> > > -----
> > > Hannu
> > No.
> > The User should be connected to the running db without restrictions
> > while backup is in progress
>
> And what's wrong with WAL-based backup then ?
>
> Especially the variant described in 23.4.4 wich keeps the WAL copied to
> backup machine up to last second ?

I think the OP means
- synchronous replication
- ability to access the standby node for queries

Yes, both requests are planned for later releases.



Bruce, can we edit the TODO to include a section called "High
Availability"?

We already have a few scattered references to such things, but in Admin
and WAL. We need a specific section.

We currently have these items, that can be moved to the new section:
- Allow a warm standby system to also allow read-only statements
- Allow WAL traffic to be streamed to another server for stand-by
replication

new section would look like this:

High Availability
-----------------
- Allow a warm standby system to also allow read-only statements
- Allow WAL traffic to be streamed to another server for stand-by
replication (synchronous/asynchronous options)
- Improve scalability of recovery for large I/O bound databases

--
Simon Riggs
2ndQuadrant http://www.2ndQuadrant.com


---------------------------(end of broadcast)---------------------------
TIP 1: if posting/reading through Usenet, please send an appropriate
subscribe-nomail command to majordomo@postgresql.org so that your
message can get through to the mailing list cleanly