[pgpool-hackers: 2146] Re: [pgpool-committers: 3836] pgpool: Branch refs/heads/bug271 was created

Sergey Kim skim at odin.com
Wed Mar 15 08:14:34 JST 2017


Sorry,

this this time I dont understand you. We have no any problems with the parameter child_max_connections!


My email was regarding the current problem #271. I want to say that in the latest patch we still could reproduce the the problem. And I sent you self contained test - extended modified test.

It seems we misunderstand each other )

________________________________
From: Tatsuo Ishii <ishii at sraoss.co.jp>
Sent: Wednesday, March 15, 2017 2:10 AM
To: Sergey Kim
Cc: pgpool-hackers at pgpool.net; Boris Voskanov
Subject: Re: [pgpool-committers: 3836] pgpool: Branch refs/heads/bug271 was created

> Yes, this value is in place.
>
> But does it make sense when comes to the being resolved issue?

Sorry, I don't understand you. This behavior regarding
child_max_connections has not been changed since at least 3.4.

Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese:http://www.sraoss.co.jp

> ________________________________
> From: Tatsuo Ishii <ishii at sraoss.co.jp>
> Sent: Wednesday, March 15, 2017 1:40:21 AM
> To: Sergey Kim
> Cc: pgpool-hackers at pgpool.net; Boris Voskanov
> Subject: Re: [pgpool-committers: 3836] pgpool: Branch refs/heads/bug271 was created
>
> Do you set child_max_connections to 1? At least the log said so. A
> Pgpool-II child process exited because of this, i.e. it seems to be a
> normal behavior.
>
>> 2017-03-14T16:06:37.894944+03:00 OAA-54999c4d7735 pgpool[25595]: [93-1] 2017-03-14 16:06:37: pid 25595:LOG:  child exiting, 1 connections reached
>> 2017-03-14T16:06:37.895079+03:00 OAA-54999c4d7735 pgpool[25595]: [93-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  child.c:406
>
> Best regards,
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese:http://www.sraoss.co.jp
>
>> Tatsuo,
>>
>>
>> this time all previously created tests succeeded but out application failed to start. So we started searching for the problem and could reproduce "balanced select problem". The test can run for minutes before reproduction of the problem (((
>>
>> I'm not sure if it can help but please pay attention that time lag between insert is select requests is very small.
>>
>>
>> Please see the pgpool logs:
>>
>> 2017-03-14T16:06:37.889485+03:00 OAA-54999c4d7735 pgpool[25595]: [79-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: B message
>> 2017-03-14T16:06:37.889702+03:00 OAA-54999c4d7735 pgpool[25595]: [79-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.889924+03:00 OAA-54999c4d7735 pgpool[25595]: [80-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 1 backend pid: 27691 statement: B message
>> 2017-03-14T16:06:37.890090+03:00 OAA-54999c4d7735 pgpool[25595]: [80-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.890248+03:00 OAA-54999c4d7735 pgpool[25595]: [81-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: Execute: BEGIN
>> 2017-03-14T16:06:37.890378+03:00 OAA-54999c4d7735 pgpool[25595]: [81-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.890535+03:00 OAA-54999c4d7735 pgpool[25595]: [82-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 1 backend pid: 27691 statement: Execute: BEGIN
>> 2017-03-14T16:06:37.890693+03:00 OAA-54999c4d7735 pgpool[25595]: [82-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.890863+03:00 OAA-54999c4d7735 pgpool[25595]: [83-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: Parse: insert into write_table(id) values($1)
>> 2017-03-14T16:06:37.891144+03:00 OAA-54999c4d7735 pgpool[25595]: [83-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.891390+03:00 OAA-54999c4d7735 pgpool[25595]: [84-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: B message
>> 2017-03-14T16:06:37.891637+03:00 OAA-54999c4d7735 pgpool[25595]: [84-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.891895+03:00 OAA-54999c4d7735 pgpool[25595]: [85-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: D message
>> 2017-03-14T16:06:37.892091+03:00 OAA-54999c4d7735 pgpool[25595]: [85-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.892272+03:00 OAA-54999c4d7735 pgpool[25595]: [86-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: Execute: insert into write_table(id) values($1)
>> 2017-03-14T16:06:37.892442+03:00 OAA-54999c4d7735 pgpool[25595]: [86-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.892642+03:00 OAA-54999c4d7735 pgpool[25595]: [87-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 1 backend pid: 27691 statement: B message
>> 2017-03-14T16:06:37.892828+03:00 OAA-54999c4d7735 pgpool[25595]: [87-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.892998+03:00 OAA-54999c4d7735 pgpool[25595]: [88-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 1 backend pid: 27691 statement: Execute: select id from write_table where id = $1
>> 2017-03-14T16:06:37.893187+03:00 OAA-54999c4d7735 pgpool[25595]: [88-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.893359+03:00 OAA-54999c4d7735 pgpool[25595]: [89-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 1 backend pid: 27691 statement: ABORT
>> 2017-03-14T16:06:37.893584+03:00 OAA-54999c4d7735 pgpool[25595]: [89-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.893839+03:00 OAA-54999c4d7735 pgpool[25595]: [90-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: ABORT
>> 2017-03-14T16:06:37.894039+03:00 OAA-54999c4d7735 pgpool[25595]: [90-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.894220+03:00 OAA-54999c4d7735 pgpool[25595]: [91-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 0 backend pid: 25885 statement: DISCARD ALL
>> 2017-03-14T16:06:37.894403+03:00 OAA-54999c4d7735 pgpool[25595]: [91-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.894587+03:00 OAA-54999c4d7735 pgpool[25595]: [92-1] 2017-03-14 16:06:37: pid 25595:LOG:  DB node id: 1 backend pid: 27691 statement: DISCARD ALL
>> 2017-03-14T16:06:37.894753+03:00 OAA-54999c4d7735 pgpool[25595]: [92-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  pool_proto_modules.c:3208
>> 2017-03-14T16:06:37.894944+03:00 OAA-54999c4d7735 pgpool[25595]: [93-1] 2017-03-14 16:06:37: pid 25595:LOG:  child exiting, 1 connections reached
>> 2017-03-14T16:06:37.895079+03:00 OAA-54999c4d7735 pgpool[25595]: [93-2] 2017-03-14 16:06:37: pid 25595:LOCATION:  child.c:406
>> 2017-03-14T16:06:37.895371+03:00 OAA-54999c4d7735 pgpool[25566]: [47-1] 2017-03-14 16:06:37: pid 25566:LOG:  child process with pid: 25595 exits with status 256
>> 2017-03-14T16:06:37.895635+03:00 OAA-54999c4d7735 pgpool[25566]: [47-2] 2017-03-14 16:06:37: pid 25566:LOCATION:  pgpool_main.c:2380
>> 2017-03-14T16:06:37.895943+03:00 OAA-54999c4d7735 pgpool[25566]: [48-1] 2017-03-14 16:06:37: pid 25566:LOG:  fork a new child process with pid: 25887
>> 2017-03-14T16:06:37.896174+03:00 OAA-54999c4d7735 pgpool[25566]: [48-2] 2017-03-14 16:06:37: pid 25566:LOCATION:  pgpool_main.c:2466
>>
>>
>>
>> And java client logs:
>> 16:06:37.890 (2) simple execute, handler=org.postgresql.jdbc.PgStatement$StatementResultHandler at 689604d9, maxRows=0, fetchSize=0, flags=1
>> 16:06:37.890 (2)  FE=> Parse(stmt=null,query="BEGIN",oids={})
>> 16:06:37.890 (2)  FE=> Bind(stmt=null,portal=null)
>> 16:06:37.890 (2)  FE=> Execute(portal=null,limit=0)
>> 16:06:37.890 (2)  FE=> Parse(stmt=null,query="insert into write_table(id) values($1)",oids={23})
>> 16:06:37.890 (2)  FE=> Bind(stmt=null,portal=null,$1=<-613120604>)
>> 16:06:37.890 (2)  FE=> Describe(portal=null)
>> 16:06:37.890 (2)  FE=> Execute(portal=null,limit=0)
>> 16:06:37.890 (2)  FE=> Sync
>> 16:06:37.892 (2)  <=BE ParseComplete [null]
>> 16:06:37.892 (2)  <=BE BindComplete [unnamed]
>> 16:06:37.892 (2)  <=BE CommandStatus(BEGIN)
>> 16:06:37.892 (2)  <=BE ParseComplete [null]
>> 16:06:37.892 (2)  <=BE BindComplete [unnamed]
>> 16:06:37.892 (2)  <=BE NoData
>> 16:06:37.892 (2)  <=BE CommandStatus(INSERT 0 1)
>> 16:06:37.893 (2)  <=BE ReadyForQuery(T)
>> 16:06:37.893 (2) simple execute, handler=org.postgresql.jdbc.PgStatement$StatementResultHandler at 18078bef, maxRows=0, fetchSize=0, flags=0
>> 16:06:37.893 (2)  FE=> Bind(stmt=S_1,portal=null,$1=<-613120604>)
>> 16:06:37.893 (2)  FE=> Execute(portal=null,limit=0)
>> 16:06:37.893 (2)  FE=> Sync
>> 16:06:37.895 (2)  <=BE BindComplete [unnamed]
>> 16:06:37.895 (2)  <=BE CommandStatus(SELECT 0)
>> 16:06:37.895 (2)  <=BE ReadyForQuery(T)
>> 16:06:37.895 (2)  FE=> Terminate
>> Exception in thread "main" java.lang.RuntimeException: Balanced
>> at com.pgpool.tests.Issue271_SingleThread.runTest(Issue271_SingleThread.java:57)
>> at com.pgpool.tests.TestStarter.main(TestStarter.java:37)
>>
>>
>> The java test is attached:
>>
>> P/S:
>> Pgpool was build from branch bug271:
>> [root at OAA-54999c4d7735 pgpool2]# git log -2
>> commit 0955a22fb7f30c5b9c1fabe2873c09246453ee50
>> Author: Tatsuo Ishii <ishii at postgresql.org>
>> Date:   Tue Mar 14 19:03:02 2017 +0900
>>
>>     Fix problem described in [pgpool-hackers: 2125].
>>
>>     Two problems are fixed:
>>
>>     - The writing_transaction flag is not reset at transaction
>>       committed/aborted. Before it was done in CommandComplete but now it
>>       is possible that we need to check the flag before a command complete
>>       message arrives. So at Execute, it is taken care by
>>       handle_query_context() which used to be a static function in
>>       CommandComplete.c. Also the session context memory is always cleared
>>       before start a session context.
>>
>>     - Ready for query Re-sync code path in read_kind_from_backend did not
>>       work. The pending message corresponding to the ready for query
>>       message could be a sync message, which does not have query
>>       context. But before we checked if there's a query context
>>       exists. The check is bogus and removed.
>>
>> commit 7c955b2a2dfb77623055b5eae9afe041de3105ee
>> Author: Tatsuo Ishii <ishii at postgresql.org>
>> Date:   Sun Mar 12 10:19:34 2017 +0900
>>
>>     Downgrade 1 more ereport from LOG to DEBUG1.
>>
>>
>>
>>
>> Sergey
>> ________________________________
>> From: Tatsuo Ishii <ishii at sraoss.co.jp>
>> Sent: Tuesday, March 14, 2017 1:28 PM
>> To: Sergey Kim
>> Cc: pgpool-hackers at pgpool.net
>> Subject: Re: [pgpool-committers: 3836] pgpool: Branch refs/heads/bug271 was created
>>
>> Hi Sergey,
>>
>> I have created a new branch named "bug271" which has been used to
>> develop the patch for bug 271 on private git repository. Now you could
>> use either the email attached patches or the branch if it's
>> convieninet for you. The branch was created against the master
>> branch. I will occasionaly rebase it with the master branch. In the
>> near future the branch will be merged into the master branch (and I
>> plan to back-patch it to 3.6 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
>>
>> From: pgpool-committers-bounces at pgpool.net
>> Subject: [pgpool-committers: 3836] pgpool: Branch refs/heads/bug271 was created
>> Date: Tue, 14 Mar 2017 10:19:33 +0000
>> Message-ID: <E1cnjYD-0005zz-GZ at gothos.postgresql.org>
>>
>>> Branch refs/heads/bug271 was created.
>>>
>>> View: http://git.postgresql.org/gitweb?p=pgpool2.git;a=shortlog;h=refs/heads/bug271
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-hackers/attachments/20170314/3378935e/attachment-0001.html>


More information about the pgpool-hackers mailing list