[pgpool-general: 3496] pgpool failover problems

[Steve Kuekes]

I am running pgpool version pgpool-II version 3.3.2 (tokakiboshi) in 
replication mode with 2 back ends with load balancing.  We have a very 
busy database and we cannot seem to keep both back ends up for longer 
that 12-16 hours of production.

These are the messages I get in the pgpool log when the failover occurs

Mar  5 12:50:03 ppasv57 pgpool[1421]: pgpool detected difference of the 
number of inserted, updated or deleted tuples. Possible last query was: 
"update ppadata.notebase set ntbmsgcount=$1,ntbupdated=$2 where ntbid=$3"
Mar  5 12:50:03 ppasv57 pgpool[1421]: CommandComplete: Number of 
affected tuples are: 1 0
Mar  5 12:50:03 ppasv57 pgpool[1421]: ReadyForQuery: Degenerate backends: 1
Mar  5 12:50:03 ppasv57 pgpool[1421]: ReadyForQuery: Number of affected 
tuples are: 1 0
Mar  5 12:50:03 ppasv57 pgpool[1421]: degenerate_backend_set: 1 fail 
over request from pid 1421
Mar  5 12:50:03 ppasv57 pgpool[2896]: starting degeneration. shutdown 
host ppasv59g(9999)
Mar  5 12:50:03 ppasv57 pgpool[2896]: Restart all children

This update statement is being done within a postgres transaction and it 
appears after the failover the transaction was successfully rolled back 
on both back ends.

We have the following options set in the pgpool.conf

replication_stop_on_mismatch = on
failover_if_affected_tuples_mismatch = on

Don't we need to have these options set to make sure that both back ends 
have exactly the same data?

What do I need to do to keep both back ends up?

Thanks for any help

Steve

-- 
Steve Kuekes

Physicians Pharmacy Alliance
118 MacKenan Drive, Suite 200
Cary, NC  27511
919-465-5801 direct
919-463-5555 main
919-463-5566 fax

steve.kuekes at physicianspharmacy.com