[Pgpool-general] Pgpool-II - replication problems

[Tatsuo Ishii]

> I was able to track down my locking issues to a few really slow
> transactions. As for the client idle time recovery it seems to work fine in
> my tests, but when I used it in production it didn't for some reason.
> 
> Does it disconnect the clients when they become idle for X seconds or do all
> the connections have to be idle?

If you mean "conection" be the one from client to pgpool, then answer
is, it disconnects the clients when they become idle for X seconds.
--
Tatsuo Ishii
SRA OSS, Inc. Japan

> Thanks,
> Tiago
> 
> On Sat, Nov 15, 2008 at 8:10 AM, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
> 
> > > I also started to have a lot of issues with the replication mode, namely
> > > locks and connection drops that force me to restart pgpool-ii. I'll look
> > > into this further next week, but it appears there is some issue with the
> > > insert_lock functionality. Is this a known issue with 2.1 or CVS HEAD?
> >
> > No.
> >
> > > Also the client idle time in recovery functionality appears to have some
> > > issues as well. Do you want me to send you logs?
> >
> > Yes, please.
> > --
> > Tatsuo Ishii
> > SRA OSS, Inc. Japan
> >
> > > Thanks,
> > > Tiago Macedo
> > >
> > > On Fri, Nov 14, 2008 at 5:45 AM, nir osovsky <nosovsky1 at gmail.com>
> > wrote:
> > >
> > > > Hi Tatsuo,
> > > >
> > > > Please help.
> > > >
> > > > Thanks,
> > > > Nir
> > > >
> > > >
> > > > On Thu, Nov 13, 2008 at 12:28 AM, nir osovsky <nosovsky1 at gmail.com>
> > wrote:
> > > >
> > > >> Thank you, Tstuso,
> > > >> I was able to download and install the new tar.
> > > >>
> > > >> Test 1:
> > > >> =====
> > > >> The following command is now *working* and both databases updated with
> > > >> the new insert.
> > > >> psql -p 9999 -c "INSERT INTO branches  VALUES (7,5,'yesh')" -d
> > > >> bench_replication
> > > >>
> > > >> Test 2:
> > > >> =====
> > > >> The following command is working on Dev1 only but the new insert is
> > *not
> > > >> replicate* to Dev2:
> > > >> psql -h dev1 bench_replication postgres
> > > >> bench_replication=# insert into branches values (9,4,'test');
> > > >>
> > > >> How can I make it work automatically, meaning every time an
> > > >> insert,update,delete occurs in Dev1 it will replicate the same to
> > Dev2?
> > > >>
> > > >> Thanks,
> > > >> Nir
> > > >>
> > > >>
> > > >> On Wed, Nov 12, 2008 at 5:49 AM, Tatsuo Ishii <ishii at sraoss.co.jp>
> > wrote:
> > > >>
> > > >>> Thanks. I think I have figured out what's going on. pgpool tries to
> > > >>> start a new transaction if DML (for example INSERT) is sent. The
> > > >>> particular setting you use with PostgreSQL generates lots of DEBUG
> > and
> > > >>> NOTICE messages while processing BEGIN and they make pgpool
> > > >>> confused. I have committed a fix for this. The new tar ball was put
> > on
> > > >>> the same location as before. Can you please grab and try it again?
> > > >>> --
> > > >>> Tatsuo Ishii
> > > >>> SRA OSS, Inc. Japan
> > > >>>
> > > >>> > Postgresql.conf it attached
> > > >>> >
> > > >>> > Thank You!
> > > >>> >
> > > >>> > Nir
> > > >>> >
> > > >>> > On Wed, Nov 12, 2008 at 12:51 AM, Tatsuo Ishii <ishii at sraoss.co.jp
> > >
> > > >>> wrote:
> > > >>> >
> > > >>> > > Thanks for the feedback. Could you show me postgresql.conf?
> > > >>> > > --
> > > >>> > > Tatsuo Ishii
> > > >>> > > SRA OSS, Inc. Japan
> > > >>> > >
> > > >>> > > > Hi Tatsuo,
> > > >>> > > >
> > > >>> > > > THANK YOU!
> > > >>> > > > I was able to download the file, and install it (configure,
> > make &
> > > >>> make
> > > >>> > > > install).
> > > >>> > > >
> > > >>> > > > I run the command again and here are the log file you asked
> > for:
> > > >>> > > >
> > > >>> > > >  psql -p 9999 -c "INSERT INTO branches  VALUES (3,1,' ')"
> > > >>> > > > ============================================
> > > >>> > > > DEBUG:  InitPostgres
> > > >>> > > > DEBUG:  StartTransaction
> > > >>> > > > DEBUG:  name: unnamed; blockState:       DEFAULT; state:
> > INPROGR,
> > > >>> > > > xid/subid/cid: 28279195/1/0, nestlvl: 1, children: <>
> > > >>> > > > DEBUG:  CommitTransaction
> > > >>> > > > DEBUG:  name: unnamed; blockState:       STARTED; state:
> > INPROGR,
> > > >>> > > > xid/subid/cid: 28279195/1/0, nestlvl: 1, children: <>
> > > >>> > > > server closed the connection unexpectedly
> > > >>> > > >         This probably means the server terminated abnormally
> > > >>> > > >         before or while processing the request.
> > > >>> > > > connection to server was lost
> > > >>> > > >
> > > >>> > >
> > > >>>
> > > >>
> > > >>
> > > >
> > > > _______________________________________________
> > > > Pgpool-general mailing list
> > > > Pgpool-general at pgfoundry.org
> > > > http://pgfoundry.org/mailman/listinfo/pgpool-general
> > > >
> > > >
> >