Re: [HACKERS] increasing the default WAL segment size

Поиск
Список
Период
Сортировка
От tushar
Тема Re: [HACKERS] increasing the default WAL segment size
Дата
Msg-id 54be0042-96bd-5286-49c6-750fe3401828@enterprisedb.com
обсуждение исходный текст
Ответ на Re: [HACKERS] increasing the default WAL segment size  (Beena Emerson <memissemerson@gmail.com>)
Ответы Re: [HACKERS] increasing the default WAL segment size  (Beena Emerson <memissemerson@gmail.com>)
Список pgsql-hackers
On 07/06/2017 12:04 PM, Beena Emerson wrote:
> The 04-initdb-walsegsize_v2.patch has the following improvements:
> - Rebased over new 03 patch
> - Pass the wal-segsize intidb option as command-line option rathern
> than in an environment variable.
> - Since new function check_wal_size had only had two checks and was
> sed once, moved the code to ReadControlFile where it is used and
> removed this function.
> - improve comments and add validations where required.
> - Use DEFAULT_XLOG_SEG_SIZE to set the min_wal_size and
> max_wal_size,instead of the value 16.
> - Use XLogSegMaxSize and XLogSegMinSize to calculate the range of guc
> wal_segment_size instead 16 - INT_MAX.
Thanks Beena. I tested with below following scenarios  and all are 
working as expected
.)Different WAL segment size i.e 1,2,8,16,32,64,512,1024 at the time of 
initdb
.)SR setup in place.
.)Combinations of  max/min_wal_size in postgresql.conf with different 
wal_segment_size.
.)shutdown the server forcefully (kill -9) / promote slave / to make 
sure -recovery happened successfully.
.)with different utilities like pg_resetwal/pg_upgrade/pg_waldump
.)running pg_bench for substantial workloads (~ 1 hour)
.)WAL segment size is not default (changed at the time of ./configure) + 
different wal_segment_size (at the time of initdb) .

Things looks fine.

-- 
regards,tushar
EnterpriseDB  https://www.enterprisedb.com/
The Enterprise PostgreSQL Company




В списке pgsql-hackers по дате отправления:

Предыдущее
От: Dean Rasheed
Дата:
Сообщение: Re: [HACKERS] Multi column range partition table
Следующее
От: Kyotaro HORIGUCHI
Дата:
Сообщение: Re: [HACKERS] hash index on unlogged tables doesn't behave asexpected