[pgpool-general: 1721] Re: Problems running jdbc testsuite

[Tatsuo Ishii]

> Re: Tatsuo Ishii 2013-05-12 <20130513.090827.1288036071461111568.t-ishii at sraoss.co.jp>
>> One thing I noticed was in the default pgpool.conf.sample, there is a line
>> 
>> backend_hostname1 = 'host2'
>> 
>> This makes pgpool do fail over if there's no host "host2" and you have
>> only one valid backend. I just commented out the line and everything
>> seems works ok.  I'm going to comment out the line above in git
>> repository.
> 
> I digged a bit more into this, and the reason seems to be that I
> didn't have /usr/share/postgresql/9.1/contrib/pgpool-regclass.sql
> loaded into the database. With that, the tests run through, otherwise
> it gets stuck in this query:
> 
> SELECT count(*) FROM pg_catalog.pg_proc AS p WHERE
> p.proname = 'pgpool_regclass' AND has_function_privilege('postgres',
> 'pgpool_regclass(cstring)', 'execute')
> 
> ... which returns an error when run interactively, but when run from
> the test driver, it looks like the result is never fetched. (In 9.2,
> the query remains in "active" state in pg_stat_activity, but doesn't
> consume any CPU anymore.)
> 
> This seems to be a change from the previous pgpool version where I
> didn't load the file either, but the tests ran through.
> 
> I'll add the loading of the file to my test script, but it might make
> sense if someone looked into why this isn't failing with a meaningful
> error message.

Thanks for the problem analyze. Also It's strange because if
pgpool_regclass does not exist, pgpool should be smart enough to avoid
calling pgpool_regclass. However there's something wrong according to
your report. I will look into this more.
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp