[pgpool-general: 1138] Re: Aborts when reach num_init_children

Lonni J Friedman netllama at gmail.com
Sat Oct 27 03:52:43 JST 2012


You didn't say which version of pgpool you're using, which version of
postgres, or what you set max_connections to in the database.  Anyway,
is pgbench releasing these connections after it finishes doing
whatever its doing?  Is it even finishing before the 32 connections
are all in use?

On Fri, Oct 26, 2012 at 11:41 AM, Emanuel Calvo <postgres.arg at gmail.com> wrote:
> I have a script that loop in a sequence from 1 to x number of
> connections for pgbench. When reachs 33 (num_init_children=32 on the
> pgpool) it starts to fail:
>
> [root at ip-10-55-83-238 ~]# time ./tests.sh
> Client 14 aborted in state 2. Probably the backend died while processing.
> Client 7 aborted in state 2. Probably the backend died while processing.
> Client 16 aborted in state 2. Probably the backend died while processing.
> Client 4 aborted in state 2. Probably the backend died while processing.
> Client 21 aborted in state 2. Probably the backend died while processing.
> Client 10 aborted in state 2: ERROR:  portal "" does not exist
> Client 17 aborted in state 2. Probably the backend died while processing.
> Client 16 aborted in state 2. Probably the backend died while processing.
> Client 1 aborted in state 2: ERROR:  unnamed prepared statement does not exist
> Client 7 aborted in state 2: ERROR:  unnamed prepared statement does not exist
> Client 12 aborted in state 2: ERROR:  unnamed prepared statement does not exist
>
>
> Then, all the processes keep the connection but idle:
>
> pgpool server:
> root     18880 18728  0 18:31 ?        00:00:01 pgpool: postgres
> postgres 10.55.83.238(58270) idle
> root     18881 18728  0 18:31 ?        00:00:01 pgpool: postgres
> postgres 10.55.83.238(58271) idle
> root     18882 18728  0 18:31 ?        00:00:01 pgpool: postgres
> postgres 10.55.83.238(58275) idle
> root     18883 18728  0 18:31 ?        00:00:01 pgpool: postgres
> postgres 10.55.83.238(58280) idle
> root     18884 18728  0 18:31 ?        00:00:00 pgpool: postgres
> postgres 10.55.83.238(58268) idle
> root     18885 18728  0 18:32 ?        00:00:00 pgpool: postgres
> postgres 10.55.83.238(58278) idle
> root     18886 18728  0 18:32 ?        00:00:00 pgpool: postgres
> postgres 10.55.83.238(58291) idle
> root     18887 18728  0 18:32 ?        00:00:00 pgpool: postgres
> postgres 10.55.83.238(58292) idle
> root     18888 18728  0 18:32 ?        00:00:00 pgpool: postgres
> postgres 10.55.83.238(58264) idle
> root     18889 18728  0 18:32 ?        00:00:00 pgpool: postgres
> postgres 10.55.83.238(58289) idle
>
>
> master:
> 19386 ?        Ss     0:00  \_ postgres: postgres postgres
> 46.51.180.111(34558) idle
> 19387 ?        Ss     0:00  \_ postgres: postgres postgres
> 46.51.180.111(34561) idle
> 19388 ?        Ss     0:00  \_ postgres: postgres postgres
> 46.51.180.111(34564) idle
> 19389 ?        Ss     0:00  \_ postgres: postgres postgres
> 46.51.180.111(34567) idle
>
>
> If I try to connect to pgpool, obviously the server directly doesn't respond.
>
> I can provide later some debug output, but I need to know if I'm not
> the only one.


More information about the pgpool-general mailing list