[prev in list] [next in list] [prev in thread] [next in thread]
List: slony1-general
Subject: Re: [Slony1-general] Postgres Upgrade from 8.0.3 to 8.1.3 is SLOW
From: Jacques Caron <jc () oxado ! com>
Date: 2006-04-21 17:41:51
Message-ID: 7.0.0.16.0.20060421193512.064f4848 () imfeurope ! com
[Download RAW message or body]
Hi,
At 22:52 20/04/2006, Pallav Kalva wrote:
>NOTICE: truncate of "listing"."address" failed - doing delete
[...]
>This is still trying to sync up listing.address table from past 4 hours,
>this table has around 5 million records.
>
> From the log it tells me that it is either truncating/deleting each
>and every table and inserting data back into the tables using the COPY
>data from the Master node.
>
> Is this the expected behavior ?
Yes.
> I was under the impression that it
>would just insert the Delta (one week data from) records for each table,
>instead of deleting and reinserting the whole data ?
How would it now the start point of the delta?
> If this is the case I dont know how many days/weeks/months it would
>take to catch up.
Actually, the whole COPY process is not that slow, though starting
with an empty schema is definitely a lot better, especially when it
can't truncate (since the whole initial replication is done in a
transaction, there can be no VACUUM between the DELETE and the COPY,
so the table ends up being twice as big as it should be...).
Note also that there is a tiny bug in 1.1.5 that prevents the delayed
index reconstruction from being used when it needs to delete instead
of truncating the tables. You might want to take a look at:
http://gborg.postgresql.org/pipermail/slony1-general/2006-January/003729.html
Hope that helps,
Jacques.
[prev in list] [next in list] [prev in thread] [next in thread]
Configure |
About |
News |
Add a list |
Sponsored by KoreLogic