Re: pgpool ABORT + no transaction warning - Mailing list pgsql-general

On Apr 3, 2006, at 3:05 AM, Tatsuo Ishii wrote:

> The problem is not reproduced here. Do you have any idea how to
> reproduce it?

If I did, I'd resolve it. :P

I suppose I can try restarting the connection pools on this server
just to see what happens. If I am able to do that, I will post the
results.

--
Thomas F. O'Connell
Database Architecture and Programming
Co-Founder
Sitening, LLC

http://www.sitening.com/
3004 B Poston Avenue
Nashville, TN 37203-1314
615-260-0005 (cell)
615-469-5150 (office)
615-469-5151 (fax)


>> On Apr 1, 2006, at 8:01 AM, Tatsuo Ishii wrote:
>>
>>>> A while back, I posted about seeing a number of warnings from
>>>> pgpool:
>>>>
>>>> http://archives.postgresql.org/pgsql-admin/2005-03/msg00305.php
>>>>
>>>> A typical pair of statements in my postgres logs looks like this:
>>>>
>>>> WARNING:  there is no transaction in progress
>>>> STATEMENT:  ABORT
>>>>
>>>> Tatsuo Ishii declared that these were harmless, and I took (and
>>>> still
>>>> take) his word for it.
>>>>
>>>> At some point in the past with my old configuration (postgres
>>>> 8.1.3 /
>>>> pgpool 2.5.1) I did something that prevented the warnings from
>>>> showing up. On a system running postgres 8.1.3, I recently upgraded
>>>> pgpool from 2.5.1 to 3.0.1. When I upgraded pgpool, I began seeing
>>>> the warnings again after a long window of not having seen them. My
>>>> configuration files for pgpool and postgres didn't change during
>>>> the
>>>> upgrade, so I'm wondering what I might've been doing to avoid
>>>> polluting my logs with them previously and why they are showing up
>>>> again.
>>>
>>> Could you tell me how to reproduce the problem?
>>
>> Here's what happened: I performed an on-line backup and recovery to
>> move postgres from one server to another. On the new server, I
>> installed pgpool 3.0.1 and copied over my old configuration files.
>>
>> I have two instances of pgpool running on the server and am only
>> using pgpool for connection pooling, not load balancing or
>> replication. The only three settings that are different between the
>> two configurations are port, logdir, and num_init_children.
>>
>> Below is one pgpool.conf (with a few paths changed back to the
>> defaults just for obfuscation), which doesn't include
>> child_max_connections, insert_lock, or ignore_leading_white_space
>> because I just used my 2.5.1 configuration file. I assume it would
>> just  use defaults for these values if they weren't specified in the
>> file?
>>
>> listen_addresses = '*'
>> port = 9999
>> socket_dir = '/tmp'
>> backend_host_name = ''
>> backend_port = 5432
>> backend_socket_dir = '/tmp'
>> secondary_backend_host_name = ''
>> secondary_backend_port = 0
>> num_init_children = 64
>> max_pool = 4
>> child_life_time = 300
>> connection_life_time = 10
>> logdir = '/tmp/pgpool1'
>> replication_mode = false
>> replication_strict = true
>> replication_timeout = 5000
>> load_balance_mode = false
>> weight_master = 0.5
>> weight_secondary = 0.5
>> replication_stop_on_mismatch = false
>> reset_query_list = 'ABORT; RESET ALL; SET SESSION AUTHORIZATION
>> DEFAULT'
>> print_timestamp = true
>> master_slave_mode = false
>> connection_cache = true
>> health_check_timeout = 20
>> health_check_period = 0
>> health_check_user = 'nobody'