Hello Andrew, hello List,
I made several tests this morning to understand perfectly the slon
processes taking into account what you tell me but I still have questions.
On the master server, for which the IP address is 1.2.3.4, I launched
this morning the "slon_start node1" command and with the "ps" command, I
saw :
-bash-3.00$ slon_start node1
Invoke slon for node 1 - /usr/bin//slon -s 1000 -d2 replication
'host=1.2.3.4 dbname=cave user=slony port=5432' 2>&1 >
/SLONY_PGS/log/slony1/slony1/node1/cave-2006-10-17_08:55:21.log &
Slon successfully started for cluster replication, node node1
PID [21865]
Start the watchdog process as well...
-bash-3.00$ ps -ef | grep post
postgres 14872 1 0 08:50 pts/1 00:00:00 /usr/bin/postmaster
postgres 14873 14872 0 08:50 pts/1 00:00:00 postgres: logger process
postgres 14895 14872 0 08:50 pts/1 00:00:00 postgres: writer process
postgres 14896 14872 0 08:50 pts/1 00:00:00 postgres: stats buffer
process
postgres 14897 14896 0 08:50 pts/1 00:00:00 postgres: stats
collector process
postgres 21865 1 0 08:55 pts/1 00:00:00 /usr/bin//slon -s 1000
-d2 replication host=1.2.3.4 dbname=cave user=slony port=5432
postgres 21875 1 1 08:55 pts/1 00:00:00 /usr/bin/perl
/usr/local/slony/bin/slon_watchdog
--config=/usr/local/slony/etc/slon_tools.conf node1 30
postgres 21876 21865 0 08:55 pts/1 00:00:00 /usr/bin//slon -s 1000
-d2 replication host=1.2.3.4 dbname=cave user=slony port=5432
postgres 21879 14872 0 08:55 pts/1 00:00:00 postgres: slony cave
1.2.3.4(35829) idle
postgres 21884 14872 0 08:55 pts/1 00:00:00 postgres: slony cave
1.2.3.4(35830) idle
postgres 21885 14872 0 08:55 pts/1 00:00:00 postgres: slony cave
1.2.3.4(35831) idle
postgres 21886 14872 0 08:55 pts/1 00:00:00 postgres: slony cave
1.2.3.4(35832) idle
=> I still don't understand why I see all these connections on the
database "cave"...
=> Could you explain me this point please ?
Then, I checked my configuration with :
-bash-3.00$ show_configuration
Slony Configuration
-------------------------------------
Slony-I Cluster: replication
Logs stored under /SLONY_PGS/log/slony1
Slony Binaries in: /usr/bin/
Node information
--------------------------------
Node: 1 Host: 1.2.3.4 User: slony Port: 5432 Forwarding?
Parent: 0 Database: cave
DSN: host=1.2.3.4 dbname=cave user=slony port=5432
Node: 2 Host: 5.6.7.8 User: slony Port: 5432 Forwarding?
Parent: 0 Database: cave
DSN: host=5.6.7.8 dbname=cave user=slony port=5432
By launching the "slon_start" command on the master node, I saw that in
the slave node a connection on the database "cave" used in the
replication from the master node :
[EMAIL PROTECTED] ~]$ ps -ef | grep post
postgres 22970 1 0 Oct12 ? 00:00:00
/opt/postgre/V813/bin/postmaster
postgres 22998 22970 0 Oct12 ? 00:00:00 postgres: writer
process
postgres 22999 22970 0 Oct12 ? 00:00:00 postgres: stats buffer
process
postgres 23000 22999 0 Oct12 ? 00:00:00 postgres: stats
collector process
postgres 21011 22970 0 08:55 ? 00:00:00 postgres: slony cave
1.2.3.4(35833) idle
Until this point, I can understand that on the slave node, I have a
connection from the master node.
But, when I launched the same command "slon_start" on the slave node, I
had :
[EMAIL PROTECTED] ~]$ ps -ef | grep post
postgres 22970 1 0 Oct12 ? 00:00:00
/opt/postgre/V813/bin/postmaster
postgres 22998 22970 0 Oct12 ? 00:00:00 postgres: writer
process
postgres 22999 22970 0 Oct12 ? 00:00:00 postgres: stats buffer
process
postgres 23000 22999 0 Oct12 ? 00:00:00 postgres: stats
collector process
postgres 21011 22970 0 08:55 ? 00:00:00 postgres: slony cave
1.2.3.4(35833) idle
root 30310 23871 0 09:01 pts/1 00:00:00 su - postgres
postgres 30312 30310 0 09:01 pts/1 00:00:00 -bash
postgres 32645 1 0 09:02 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32654 1 0 09:02 pts/1 00:00:00 /usr/bin/perl
/usr/local/slony/bin/slon_watchdog
--config=/usr/local/slony/etc/slon_tools.conf node2 30
postgres 32689 32645 0 09:02 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32692 22970 0 09:02 ? 00:00:00 postgres: slony cave
5.6.7.8(35715) idle
postgres 32699 22970 0 09:02 ? 00:00:00 postgres: slony cave
5.6.7.8(35716) idle
postgres 32700 22970 0 09:02 ? 00:00:00 postgres: slony cave
5.6.7.8(35718) idle
postgres 32701 22970 0 09:02 ? 00:00:00 postgres: slony cave
5.6.7.8(35719) idle
=> It is the same problem that I have on the master node, I don't
understand all the connections on the database "cave"...
[EMAIL PROTECTED] ~]$ show_configuration
Slony Configuration
-------------------------------------
Slony-I Cluster: replication
Logs stored under /SLONY_PGS/log/slony1
Slony Binaries in: /opt/postgre/V813/bin/
Node information
--------------------------------
Node: 1 Host: 1.2.3.4 User: slony Port: 5432 Forwarding?
Parent: 0 Database: cave
DSN: host=1.2.3.4 dbname=cave user=slony port=5432
Node: 2 Host: 5.6.7.8 User: slony Port: 5432 Forwarding?
Parent: 0 Database: cave
DSN: host=5.6.7.8 dbname=cave user=slony port=5432
To be sure, I launched the commands "ps -eLf | grep post", "ps axm |
grep post", "netstat | grep post" on the master and slave nodes, but I
didn't find a good answer...
This is the results of these commands on the MASTER node :
-bash-3.00$ ps -eLf | grep post
postgres 14872 1 14872 0 1 08:50 pts/1 00:00:00
/usr/bin/postmaster
postgres 14873 14872 14873 0 1 08:50 pts/1 00:00:00 postgres:
logger process
postgres 14895 14872 14895 0 1 08:50 pts/1 00:00:00 postgres:
writer process
postgres 14896 14872 14896 0 1 08:50 pts/1 00:00:00 postgres:
stats buffer process
postgres 14897 14896 14897 0 1 08:50 pts/1 00:00:00 postgres:
stats collector process
postgres 21865 1 21865 0 1 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21875 1 21875 0 1 08:55 pts/1 00:00:00 /usr/bin/perl
/usr/local/slony/bin/slon_watchdog
--config=/usr/local/slony/etc/slon_tools.conf node1 30
postgres 21876 21865 21876 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21876 21865 21877 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21876 21865 21878 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21876 21865 21880 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21876 21865 21881 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21876 21865 21882 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21876 21865 21883 0 7 08:55 pts/1 00:00:00
/usr/bin//slon -s 1000 -d2 replication host=1.2.3.4 dbname=cave
user=slony port=5432
postgres 21879 14872 21879 0 1 08:55 pts/1 00:00:02 postgres:
slony cave 1.2.3.4(35829) idle
postgres 21884 14872 21884 0 1 08:55 pts/1 00:00:00 postgres:
slony cave 1.2.3.4(35830) idle
postgres 21885 14872 21885 0 1 08:55 pts/1 00:00:01 postgres:
slony cave 1.2.3.4(35831) idle
postgres 21886 14872 21886 0 1 08:55 pts/1 00:00:04 postgres:
slony cave 1.2.3.4(35832) idle
postgres 1182 14872 1182 0 1 09:02 pts/1 00:00:04 postgres:
slony cave 5.6.7.8(35717) idle
-bash-3.00$ ps axm | grep post
14872 pts/1 - 0:00 /usr/bin/postmaster
14873 pts/1 - 0:00 postgres: logger process
14895 pts/1 - 0:00 postgres: writer process
14896 pts/1 - 0:00 postgres: stats buffer process
14897 pts/1 - 0:00 postgres: stats collector process
21879 pts/1 - 0:02 postgres: slony cave 1.2.3.4(35829) idle
21884 pts/1 - 0:00 postgres: slony cave 1.2.3.4(35830) idle
21885 pts/1 - 0:01 postgres: slony cave 1.2.3.4(35831) idle
21886 pts/1 - 0:04 postgres: slony cave 1.2.3.4(35832) idle
1182 pts/1 - 0:04 postgres: slony cave 5.6.7.8(35717) idle
-bash-3.00$ netstat | grep post
tcp 0 0 bt1.frec.bull.fr:35830
bt1.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:35831
bt1.frec.bull.fr:postgres ESTABLISHED
tcp 1 0 bt1.frec.bull.fr:35828
bt1.frec.bull.fr:postgres CLOSE_WAIT
tcp 0 0 bt1.frec.bull.fr:35829
bt1.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:35832
bt1.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:postgres
bt1.frec.bull.fr:35829 ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:postgres
bt1.frec.bull.fr:35831 ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:postgres
bt1.frec.bull.fr:35830 ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:postgres
bt1.frec.bull.fr:35832 ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:postgres
bt3.frec.bull.fr:35717 ESTABLISHED
tcp 0 0 bt1.frec.bull.fr:35833
bt3.frec.bull.fr:postgres ESTABLISHED
And on the SLAVE node :
[EMAIL PROTECTED] baseSlony]$ ps -eLf | grep post
postgres 22970 1 22970 0 1 Oct12 ? 00:00:00
/opt/postgre/V813/bin/postmaster
postgres 22998 22970 22998 0 1 Oct12 ? 00:00:00 postgres:
writer process
postgres 22999 22970 22999 0 1 Oct12 ? 00:00:00 postgres:
stats buffer process
postgres 23000 22999 23000 0 1 Oct12 ? 00:00:00 postgres:
stats collector process
postgres 21011 22970 21011 0 1 08:55 ? 00:00:04 postgres:
slony cave 1.2.3.4(35833) idle
postgres 32645 1 32645 0 1 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32654 1 32654 0 1 09:03 pts/1 00:00:00 /usr/bin/perl
/usr/local/slony/bin/slon_watchdog
--config=/usr/local/slony/etc/slon_tools.conf node2 30
postgres 32689 32645 32689 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32689 32645 32690 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32689 32645 32691 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32689 32645 32695 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32689 32645 32696 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32689 32645 32697 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32689 32645 32698 0 7 09:03 pts/1 00:00:00
/opt/postgre/V813/bin//slon -s 1000 -d2 replication host=5.6.7.8
dbname=cave user=slony port=5432
postgres 32692 22970 32692 0 1 09:03 ? 00:00:02 postgres:
slony cave 5.6.7.8(35715) idle
postgres 32699 22970 32699 0 1 09:03 ? 00:00:00 postgres:
slony cave 5.6.7.8(35716) idle
postgres 32700 22970 32700 0 1 09:03 ? 00:00:01 postgres:
slony cave 5.6.7.8(35718) idle
postgres 32701 22970 32701 0 1 09:03 ? 00:00:03 postgres:
slony cave 5.6.7.8(35719) idle
[EMAIL PROTECTED] baseSlony]$ ps axm | grep post
22970 ? - 0:00 /opt/postgre/V813/bin/postmaster
22998 ? - 0:00 postgres: writer process
22999 ? - 0:00 postgres: stats buffer process
23000 ? - 0:00 postgres: stats collector process
21011 ? - 0:04 postgres: slony cave 1.2.3.4(35833) idle
32645 pts/1 - 0:00 /opt/postgre/V813/bin//slon -s 1000 -d2
replication host=5.6.7.8 dbname=cave user=slony port=5432
32689 pts/1 - 0:02 /opt/postgre/V813/bin//slon -s 1000 -d2
replication host=5.6.7.8 dbname=cave user=slony port=5432
32692 ? - 0:02 postgres: slony cave 5.6.7.8(35715) idle
32699 ? - 0:00 postgres: slony cave 5.6.7.8(35716) idle
32700 ? - 0:01 postgres: slony cave 5.6.7.8(35718) idle
32701 ? - 0:03 postgres: slony cave 5.6.7.8(35719) idle
[EMAIL PROTECTED] baseSlony]$ netstat | grep post
tcp 0 0 bt3.frec.bull.fr:postgres
bt3.frec.bull.fr:35716 ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:postgres
bt3.frec.bull.fr:35719 ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:postgres
bt3.frec.bull.fr:35718 ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:postgres
bt3.frec.bull.fr:35715 ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:35718
bt3.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:35719
bt3.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:35716
bt3.frec.bull.fr:postgres ESTABLISHED
tcp 1 0 bt3.frec.bull.fr:35714
bt3.frec.bull.fr:postgres CLOSE_WAIT
tcp 0 0 bt3.frec.bull.fr:35715
bt3.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:35717
bt1.frec.bull.fr:postgres ESTABLISHED
tcp 0 0 bt3.frec.bull.fr:postgres
bt1.frec.bull.fr:35833 ESTABLISHED
I don't understand very well what happens with the "slon_start" command
and I hope someone will give me an explanation or maybe a documentation
in which I can find information about the internal process of Slony-I.
Sorry for the length of this e-mail and thank you to have read ot ! ;-)
Regards,
Alexandra
Andrew Sullivan wrote:
>On Mon, Oct 16, 2006 at 08:28:05AM -0400, Andrew Sullivan wrote:
>
>
>>On Mon, Oct 16, 2006 at 08:55:29AM +0200, DANTE Alexandra wrote:
>>
>>
>>>(very detailed, thanks !) but I didn't find a chapter or an other
>>>documentation in which all the "slon" processes created with the
>>>execution of the "slon_start" command are detailed.
>>>
>>>
>
>So, to answer the original question, the reason you see all those
>slon processes is because the replication system is multithreaded.
>On Linux, threads show up in ps as processes. I don't think you do
>have that many open postgres connections. If I'm reading your ps
>output correctly, it's only one per side, which is what I'd figure.
>You can check for sure with netstat, I suppose.
>
>A
>
>
>
_______________________________________________
Slony1-general mailing list
[email protected]
http://gborg.postgresql.org/mailman/listinfo/slony1-general
