On 7/15/2004 3:07 PM, Guthrie, Jeremy wrote:

> First off, congrats to the writers of Slony.  Excellent software.
> 
> I am putting it through its paces as I try to update my application to better account for Slony.  My application au
> tomatically takes care of weekly reindexing because of the volume of data that changes daily.  However it is runnin
> g into errors when looking at the Slony Indexes.  Reindex is erroring out on the '-'.  I thought I saw something ab
> out this before but I cannot find where.   Specifically, the index is  tsyslog__Slony-I_tsyslog_cluster_rowID_key.
> 

You would have to double quote that index name because of the mixed case 
and dash.

I wonder though if that reindexing is an artefact left over from running 
older versions of PostgreSQL. Are you sure that the version you are 
running currently still suffers from index bloat despite of frequent 
(non-full) vacuuming?

> 1.
> a)  should I worry about having to reindex this?
> b)  if so, how?  psql won't let me.
> 
> 2.  I've noticed that the above listed index seems to stay the same size after running a replication test.  Is my a
> ssumption correct that because of how Slony behaves that it will rarely leave unused index space?
> 
> 3.  By doing a full vacuum or a reindex, will that block an entire replication set or the one table of the set till
>  the full vacuum or reindex finishes?

full vacuum as well as reindex acquire an exclusive lock on the table. 
You're not doing your application a favor with doing that. You should 
rather look into pg_autovacuum or setting up your own proper vacuum 
scheduling.

> 4.  If you want me to be quiet, just say so.  ;)

Questions welcome.


Jan


-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #