[pgpool-hackers: 1792] Re: Problem with locking in In Memory Cache

[Huong Dangminh]

Ishii-san,

Many thanks for patching.
The patch work fine.

Best regards,
---
Dang Minh Huong
NEC Solution Innovators, Ltd.
http://www.nec-solutioninnovators.co.jp/en/

> -----Original Message-----
> From: Tatsuo Ishii [mailto:ishii at sraoss.co.jp]
> Sent: Tuesday, September 06, 2016 8:06 AM
> To: Dangminh Huong(ダンＭ フーン) <huo-dangminh at ys.jp.nec.com>;
> Yanagisawa Hiroshi(柳澤 博) <hir-yanagisawa at ut.jp.nec.com>;
> pgpool-hackers at pgpool.net
> Subject: Re: [pgpool-hackers: 1789] Re: Problem with locking in In
> Memory Cache
> 
> You are right. Pgpool-II should have fowarded sync message from
> frontend to backend. Attached is a patch against 3.4 stable tree.
> 
> Best regards,
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese:http://www.sraoss.co.jp
> 
> > Thank you for the report. I will look into this.
> >
> > Best regards,
> > --
> > Tatsuo Ishii
> > SRA OSS, Inc. Japan
> > English: http://www.sraoss.co.jp/index_en.php
> > Japanese:http://www.sraoss.co.jp
> >
> >> Hi,
> >>
> >> I have found a problem with table locking In In Memory Cache feature.
> >> Confirmed in pgpool-II 3.4.8 in both memcached and sharedmemory
> mode.
> >>
> >> When using JDBC to access to pgpool-II. In the second time of (the
> same) query
> >> executing, it returned result from cached, but after that
> pg_stat_activity's state of query
> >> was still in 'active' and not change.
> >> So the locks made by executed query was not release too.
> >>
> >> Take a look at source and i found that, in extended protocol, locking
> of table
> >> was made when 'P'(Parse) message is sent, and release when 'S'(Sync)
> message
> >> is sent to backend.
> >> But when memory_cache_enabled is true. Parse message is sent but
> Sync message
> >> is not sent to backend.
> >> In pool_fetch_from_memory_cache function. It sound that when cache
> found,
> >> Sync message from frontend was discard but not forward to backend
> to finish query.
> >> I think, it is necessary to tell to backend to finish executed query
> here, is that right?
> >>
> >> Best regards,
> >> ---
> >> Dang Minh Huong
> >> NEC Solution Innovators, Ltd.
> >> http://www.nec-solutioninnovators.co.jp/en/
> >> _______________________________________________
> >> pgpool-hackers mailing list
> >> pgpool-hackers at pgpool.net
> >> http://www.pgpool.net/mailman/listinfo/pgpool-hackers
> > _______________________________________________
> > pgpool-hackers mailing list
> > pgpool-hackers at pgpool.net
> > http://www.pgpool.net/mailman/listinfo/pgpool-hackers