What I always do is "ps -axf", which will display a "tree" of the processes : you'll know which ones are children of
theothers.
I don't know if the "dead" postmaster will be shown in a readable location... The problem seems to be somewhere else
(kills/signalsto the postmaster/backends, not correctly disconnected connections, etc)
Nicolas Huillard
G.H.S
Directeur Technique
Tél : +33 1 43 21 16 66
Fax : +33 1 56 54 02 18
mailto:nhuillard@ghs.fr
http://www.ghs.fr
-----Message d'origine-----
De: Len Morgan [SMTP:len-morgan@crcom.net]
Date: samedi 22 avril 2000 17:02
À: pgsql-admin@postgresql.org
Objet: [ADMIN] Dead Postmasters
On a couple of my installations, I've noticed that after a few days there
are several "dead" postmasters (i.e., not associated with any running
backend). When I do a ps -ax (RedHat 6.1, Postgresql 6.5.3) I see:
..
[postmaster]
[postmaster]
[postmaster]
[postmaster]
[postmaster]
[postmaster]
..
and several
/usr/bin/postgres ..... lines
The problem is that there are more [postmaster] lines (sometimes 4 or 5
times as many) as there are /usr/bin/postgres lines.
When I do a kill -TERM pid on all of the [postmaster] pids or restart the
postmaster, the system gets much faster. The problem is I also cut of
"live" connections in the process. My question is: Is there any way I can
determine which of the [postmaster] entries are associated with which
/usr/bin/postgres entries?
Thank you!
Len Morgan
