On Wed, 10 Jan 2007, Largo Stemper wrote:

> Hi Everydoby
> 
> 
> That s why I wanted to know generally , based on the experience of everyone ; 
> where did U found CONSTRAINTs or PROBLEMs (generally) while using SLONY in a 
> project and not in a simple DB replication. As example for what I am looking 
> for is : - SLONY-I is "good" by handling random() values because I will not 
> have different random-Vals in Master or Slave , what is not the case of 
> PGCluster. - A disadvantage, as example, can be that SLONY-I can not handle a 
> project that needs every week or day a new database because, we can not stop 
> a replication , but we must add a new set. - Another example is that one may 
> use lot of sets by a large number
> of DBs to avoid deadlocks, thing that makes the administration more 
> complicated...

I'm a bit confused as to what your issues are with adding tables.
If you want to add a table to an existing setup without stopping replication 
you would do something like:
-Create a new set
-Add your new tables to the new set
-Subscribe your slaves to the new set
-Wait till everything replicates
-Merge your new set into your old set

That leaves you with 1 set at the end of the process so it should not be an 
administration pain.  I don't belive any of the above operations would require 
an outage window and should not be painful especially if you write a script to 
automate the above steps.

What we have found much more painful is adding columns (or any other ALTER 
table) command to tables that are already being replicated.  In the current 
version of slony this requires obtain an exclusive lock on all replicated 
tables in all sets.  This essentially requires an outage/maintaince window to 
avoid deadlock.

It is unclear to me what you mean by 'one may use lot of sets by a large number 
of DB's to avoid deadlock'


The admin guide is full of lots of things that can cause problems for you if 
you don't watch out for them and is well worth your time to read in detail.
Ignoring the issues raised in the locking chapter will give you some nasty 
surprises but if you gain a good understanding of how slony behaves you 
shouldn't be surprised.  Which behaviours of slony will be big issues depends a 
lot on your operational conditions.

> 
> I mean according to your projects and your expereince, where did one of you 
> all see some problems.
> 
> 
> Thanks again
> 
> -----------------------------------------------------------------------------------------------------------------------------------
> Christopher Browne wrote:
>> Largo Stemper wrote:
>>> Can someone help me by giving me a briefing listing the disadvantages of 
>>> SLONY-I.
>>> 
>> Disadvantages (and advantages) do not exist in a vacuum.
>> 
>> They have to be framed in some context; Slony-I can only have advantages
>> or disadvantages when compared to something else.
>> 
>> You can't conceivably have either sort of "vantage" without specifying
>> that "something else."
> 
> 
> _______________________________________________________________________
> Viren-Scan f?r Ihren PC! Jetzt f?r jeden. Sofort, online und kostenlos.
> Gleich testen! http://www.pc-sicherheit.web.de/freescan/?mc=022222
> 
> _______________________________________________
> Slony1-general mailing list
> Slony1-general at gborg.postgresql.org
> http://gborg.postgresql.org/mailman/listinfo/slony1-general
>