[pgpool-hackers: 4506] Re: Proposal to backport notice_per_node_statement

[Bo Peng]

Ishii-san,

> The feature is pretty useful for testing because it allows to take
> psql output logs for queries pgpool sends to each backend node, then
> compare the logs with expected output so that we can confirm pgpool
> works as expected. This is similar to PostgreSQL's regression test. We
> maybe able to do similar thing by:
> 
> pgpool set client_min_messages to log;
> 
> But this has disadvantages:
> 
> 1) This will print internal rel cache access log which makes the
> comparison with the expected file unstable because the log is added
> depending on the pgpool relation cache.
> 
> 2) The log usually includes process id, again which makes the
> comparison unstable.
> 
> notice_per_node_statement does not have these disadvantages.
> 
> If we backport notice_per_node_statement, probably we should implement
> it as "undocumented" feature because our policy is "do not backport
> new feature to back branches". I think making it as an undocumented
> feature is a good compromize.
> 
> Comments/suggestions are welcome.

Yes. I agree to backport it and make it as an undocumented feature.

-- 
Bo Peng <pengbo at sraoss.co.jp>
SRA OSS K.K.
TEL: 03-5979-2701 FAX: 03-5979-2702
URL: https://www.sraoss.co.jp/