[Pgpool-general] replication between 32 bit and 64 bit server

[Filip Rembiałkowski]

W dniu 27 sierpnia 2009 11:46 użytkownik Asaf <asaf at lingnu.com> napisał:
>> instead of using rsync or tar,
>> use pg_dumpall (for all databases) or pg_dump (for one database) and restore the dump on second machine with psql and/or pg_restore.
>>
>> pg_dumpall  | psql -h othermachine postgres
>>
>> see Backup section in the docs: http://www.postgresql.org/docs/current/static/backup.html
>>
> I did that, replication didn't work, so I thought the pg_dumpall is not
> good enough for this,

strange, it should work. maybe try to do the test with an empty database first?


>
> when something don't work, how can I get more details from pgpool (i
> didn't see errors, and it said he had 2 backends)

try "-d" option for more detailed logging

see http://pgpool.projects.postgresql.org/contrib_docs/pgpool-II_for_beginners.pdf
for more info; there's an instruction how to enable logging...


>>
>> PS. I would rather not use pgpool for replication - it adds some limitations/gotchas (think of volatile functions like random or current_timestamp).
>> I'd use pgpool only for LB and pooling - and leave replication to slony or skytools.
>>
> I've tried slony, but it requires alot of maintainece,


ouch. why? follow these steps to setup slony-I with slon-tools:

* write the config file
* slonik_init_cluster
* slonik_create_set set1
* slon_start 1
* slon_start 2
* slonik_subscribe_set set1 2



>
> I'm looking for something completely transparent to users,
> such as mysql replication, pgpool seemed to do the job

slony-I (or other trigger-based) replication is actually more
transparent than pgpool-ii replication.




--
Filip Rembiałkowski
JID,mailto:filip.rembialkowski at gmail.com
http://filip.rembialkowski.net/