Backend process is still runs even postmaster got killed

On Fri, Dec 7, 2012 at 10:21 AM, Hari Babu <haribabu.kommi@huawei.com>wrote:

Backend processes are still running even if the postmaster got killed and
all other server processes are exited by checking the****

Postmaster status. ****

** **

And the backend process is providing the service to the client.****

Is there any problems are possible? If we continue the system for a long
time with the above behavior?****

** **

Until all these kind of backend processes are exited, postmaster is
failing to start.****

Manual cleanup is required for these backend processes? Or is there anyway
we can handle?****

** **

Regards,****

Hari babu.****

Hi Hari.

How exactly do you kill the postmaster? It is suggested you use pg_ctl for
server shutdowns/restarts. Check out the option -m:
-m SHUTDOWN-MODE can be "smart", "fast", or "immediate"
Shutdown modes are:
smart quit after all clients have disconnected
fast quit directly, with proper shutdown
immediate quit without complete shutdown; will lead to recovery on
restart

"Smart" will wait for all your backend processes to terminate (or fail if
it takes too long). If i remember correctly "fast" will cause the
postmaster to exit even if backends are still up, the equivalent of sending
a kill -15 (SIGTERM). And "immediate" is like sending a kill -9 (SIGKILL)
i.e. violent shutdown of everything.

I'm guessing you went for the equivalent of SIGTERM that's why you still
have backends running.

regards,
--thalis

On Fri, Dec 7, 2012 at 7:26 PM, Thalis Kalfigkopoulos

<tkalfigo(at)gmail(dot)com> wrote:

On Fri, Dec 7, 2012 at 10:21 AM, Hari Babu <

<mailto:haribabu.kommi@huawei.com> haribabu(dot)komm(at)huawei(dot)com>
wrote:

Backend processes are still running even if the postmaster got killed and

all other server processes are exited by checking the

Postmaster status.

And the backend process is providing the service to the client.

Is there any problems are possible? If we continue the system for a long

time with the above behavior?

Until all these kind of backend processes are exited, postmaster is

failing to start.

Manual cleanup is required for these backend processes? Or is there anyway

we can handle?

Hi Hari.

How exactly do you kill the postmaster? It is suggested you use pg_ctl for

server shutdowns/restarts. Check out the option -m:

-m SHUTDOWN-MODE can be "smart", "fast", or "immediate"
Shutdown modes are:
smart quit after all clients have disconnected
fast quit directly, with proper shutdown
immediate quit without complete shutdown; will lead to recovery on

restart

"Smart" will wait for all your backend processes to terminate (or fail if

it takes too long). If i remember correctly "fast" will cause the postmaster
to exit even if >backends are still up, the equivalent of sending a kill -15
(SIGTERM). And "immediate" is like sending a kill -9 (SIGKILL) i.e. violent
shutdown of everything.

I'm guessing you went for the equivalent of SIGTERM that's why you still

have backends running.

Usually we are using the pg_ctl with shutdown option only.

Actually I thought of asking the postmaster crash scenario, which I have
simulated with kill -9 PID.

I know postmaster crash is a very rare scenario, In this I want to know what
are all the problems can occur?

Regards,

Hari babu.

Hari Babu wrote:

Backend processes are still running even if the postmaster got killed and
all other server processes are exited by checking the

Postmaster status.

What server version?

--
Álvaro Herrera http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general

On Fri, Dec 7, 2012 at 10:21 AM, Alvaro Herrera
<alvherre(at)2ndquadrant(dot)com> wrote:

Hari Babu wrote:
Backend processes are still running even if the postmaster got killed and
all other server processes are exited by checking the

Postmaster status.

What server version?

I had tested it in 9.3 devel and 9.2 as well.
I tested as follows.

./pg_ctl -D data start
./psql postgres

In another terminal kill the postmaster with kill -9 PID.
Now start using the psql which is already connected, it can execute all
queries.
And check the server processes (ps -elf | grep postgres) only backend
process will remain.

I know postmaster crash is a very rare scenario, If backend continuously
providing the service,
I want to know what are all the problems can occur?

Regards,
Hari babu.

--
Sent via pgsql-general mailing list (pgsql-general@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general