[Pgcluster-general] pgcluster 1.5.0-rc14 problems

Tarhon-Onu Victor mituc at activemediatech.com
Fri Feb 16 11:34:54 UTC 2007


On Thu, 15 Feb 2007, a.mitani at sra-europe.com wrote:

> Would you try to set "no" to <Use_Replication_Log> , if you do not have a
> lower replication server.

 	I've found the problem, the solution was as stupid simple as the 
problem.

 	It seems like that setting has nothing to do with this. I just 
tested and it seems like the "Host_Name" value from the cluster.conf file 
on each cluster member is very important.

 	When IP addresses are used in cluster configuration you must make 
sure that the FQDN is associated with that IP address. Otherwise you must 
use FQDNs associated with the IP addresses used, of if you don't have DNS 
settings for those yet (as it was in my case, I used some IP addresses set 
on two NICs linked by a cross-over netwrk cable) you have to use as 
Host_Name that specific IP address.

 	In my setup the servers were named chip and dale. The IP addresses 
associated in my local DNS server with these names were 10.100.100.129 and 
10.100.100.130.
 	The IP addresses used for postgres replication were 10.100.103.1 
and 10.100.103.2 (/30) and were configured on a cross-over cable.
 	Since I've ignored the Host_Name setting (or ocasionally I've set 
it to the real hostname) this was by default the FQDN of the db server 
(chip or dale).
 	So, probably, for some reason, the replicator was receiving 
packets from 10.100.103.1 and hostname chip, at it was trying to get a 
status for the query from chip, which was 10.100.100.129.

 	So, problem solved, PGCluster is working great and is ready to be 
pgbench-ed :) Thanks guys!:)

P.S.: Btw, for Fedora Core users, I've created some RPM files for 
postgresql 8.1.7 and 8.1.8 (latest FC6 updates) with the pgcluster patch 
applied (i'll create some for postgresql 8.2 too). It's not a big deal to 
create those RPMs yourself, but if someone is interested just let me know 
and I'll upload them on an FTP server after I'll finish writing some 
startup scripts for the cluster and banalcer and eventually tweak the 
postgresql startup script a little bit.

-- 
Victor Tarhon-Onu
Lead Systems Administrator
ActiveMedia Technology
http://www.activemediatech.com
.............................................................................
This communication contains information that is confidential and may also be
privileged. It is for the exclusive use of the intended recipient(s). If you
are not the intended recipient(s) please note that any form of distribution,
copying or use of this communication or the information in it is strictly
prohibited and may be unlawful. If you have received this communication in
error kindly notify the sender immediately and delete it from your system.


More information about the Pgcluster-general mailing list