[Pgpool-general] ProcessFrontendResponse: failed to read kind from frontend. frontend abnormally exited

Benoit Poulet benoit.poulet at businessdecision.com
Mon Sep 24 14:37:00 UTC 2007


Hello,

I'm using Pgpool 3.4 in a production environnement with 2 postgresql
8.2.3 as backend.
The software is written in java.

-----server 1-----|-----server 2-----
JAVA APPLICATION
      |
    PGPOOL-------------POSTGRESQL
      |
  POSTGRESQL

Pgppool is set in replication mode, but NOT in load balancing.

I got a lot of problem with pgpool, a lot of database desynch, a bad
behavior when the secondary backend failed.

** 1rst problem - Databases desynch

Nearly every day, i got a diff between the 2 servers. some data are lost
on the secondary.
Some data who was correctly inserted in the primary server  are not
present on the secondary.

When i look into the log i can often see this message :
2007-09-23 22:26:53 ERROR: pid 10686: ProcessFrontendResponse: failed to
read kind from frontend. frontend abnormally exited
2007-09-23 22:26:53 LOG:   pid 10686: do_child: exits with status 1 due
to error

I would like to know if theses messages can explain the databases
desynch and if not, what can be the reason of this desynch ? (we already
check the limitation of pgpool on the pgpool's homepage).

And i would like to know too, if Pgpool is a real enterprise class
replication solution, already in production for years at some compagny
or just a experimental product.

** 2nd problem

When the secondary server failed or goes off-line, pgpool enter in
degeneration mode and get stuck !
we need to restart pgool to be able to connect to the database again.

Is it the right behavior or i missed something ?

Thanks for any help.
Regards,
Benoit.

Here you can find my pgpool.conf
########################
listen_addresses = '*'
port = 5432
socket_dir = '/tmp'
backend_host_name = ''
backend_port = 9999
backend_socket_dir = '/tmp'
secondary_backend_host_name = 'backend-02'
secondary_backend_port = 5432
num_init_children = 25
max_pool = 4
child_life_time = 0
connection_life_time = 300
child_max_connections = 0
logdir = '/var/log'
replication_mode = true
replication_strict = true
replication_timeout = 5000
load_balance_mode = false
weight_master = 0.5
weight_secondary = 0.5
replication_stop_on_mismatch = false
replicate_select = false
reset_query_list = 'ABORT; RESET ALL; SET SESSION AUTHORIZATION DEFAULT'
print_timestamp = true
master_slave_mode = false
connection_cache = true
health_check_timeout = 30
health_check_period = 60
health_check_user = 'nobody'
insert_lock = true
ignore_leading_white_space = true
log_statement = true
log_connections = false
log_hostname = true
enable_pool_hba = false
########################

-- 
*Benoit Poulet*
*Business & Decision Interactive Eolas
<http://www.businessdecision-interactive.com/?xtor=EPR-61>* - Tél :+33
(4) 76 44 50 50 - Fax : +33 (4) 76 44 00 41
Consultant Expert Web
Portails et Applications collaboratives, Hosting, Visites et
transformations ... en mode Projet complet, ASP, Centre de Services
/*Nous vous confirmons : IP pour les ERP et la Business Intelligence
<http://www.businessdecision-interactive.com/703-hebergement-d-une-plate-forme-hfm.htm>*
: Cognos, Hyperion, Oracle, SAP, ... c'est très fort .... /
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://pgfoundry.org/pipermail/pgpool-general/attachments/20070924/22cfacfa/attachment.html 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: benoit.poulet.vcf
Type: text/x-vcard
Size: 172 bytes
Desc: not available
Url : http://pgfoundry.org/pipermail/pgpool-general/attachments/20070924/22cfacfa/attachment.vcf 


More information about the Pgpool-general mailing list