I was hoping someone could help shed some light on whether dead tuples  
reduce slave
performance in terms of keeping up to date with the master.

I read Shoaib's article

http://people.planetpostgresql.org/shoaibmir/index.php?/archives/9-Schema-specific-dead-tuples-report.html#extended

and decided to take a look at our Slony schema.  We get once in a  
while lags of 30 seconds or more and wondered
if it had to do with table bloat.  We are running Slony 1.2.10 on  
PostgreSQL 8.2.4.

The results on the master seemed reasonable every time I looked, i.e.,  
routinely vacuumed.

However, on one of the slaves I saw a lot of table bloat.  Can this  
hurt performance?

tii=# select * from deadtup_report;
-[ RECORD 2 ]------+--------------------
table_name         | _tii.sl_confirm
table_len          | 11214848
tuple_count        | 7762
tuple_len          | 403624
tuple_percent      | 3.6
dead_tuple_count   | 186290
dead_tuple_len     | 9687080
dead_tuple_percent | 86.38
free_space         | 315080
free_percent       | 2.81
-[ RECORD 3 ]------+--------------------
table_name         | _tii.sl_event
table_len          | 10371072
tuple_count        | 2915
tuple_len          | 243012
tuple_percent      | 2.34
dead_tuple_count   | 70657
dead_tuple_len     | 5936368
dead_tuple_percent | 57.24
free_space         | 3640376
free_percent       | 35.1
-[ RECORD 6 ]------+--------------------
table_name         | _tii.sl_log_2
table_len          | 148004864
tuple_count        | 38796
tuple_len          | 5069174
tuple_percent      | 3.43
dead_tuple_count   | 1017334
dead_tuple_len     | 134977401
dead_tuple_percent | 91.2
free_space         | 1742612
free_percent       | 1.18
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.slony.info/pipermail/slony1-general/attachments/20080207/2d61355d/attachment.htm