Обсуждение: autovacuum: vacuum analyze vs autovacuum: vacuum

Поиск
Список
Период
Сортировка

autovacuum: vacuum analyze vs autovacuum: vacuum

От
Artem Tomyuk
Дата:
Hi, Gurus!


As i understand from documentation:

when the autovacuum: vacuum analyze will be triggered, it will delete dead tuples and than will analyze table
but
when the autovacuum: vacuum will be triggered it will only mark for deletion dead tubles.

Is it valid explanation for postgres 9.4?

Thanks

Strange behavior of the timezone

От
Vlad Alexeenkov
Дата:

Hi!

Can anyone explain this behavior?

select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 15:11:10'::timestamp at time zone 'Europe/Moscow' as t2

 t1                                       t2
=================== ===================
01.01.1900 15:11:10 01.01.1900 15:40:53
Strange behavior of the timezone




select version()

version                                                                                                 
========================================================================================================
PostgreSQL 9.6.5 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623 (Red Hat 4.8.5-11), 64-bit

Thanks

Re: Strange behavior of the timezone

От
Robert Zenz
Дата:
That is not necessarily a strange behavior, as the Timezome most likely includes
all the adjustments that were made in the period of time (leap seconds and the
like). At least as far as I know.

How does it behave if you try a more current date, like today? What is your
system/database timezone?


On 27.03.2018 11:40, Vlad Alexeenkov wrote:
> Hi!
> 
> Can anyone explain this behavior?
> 
> select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 15:11:10'::timestamp at
> time zone 'Europe/Moscow' as t2
> 
>  t1                                       t2
> =================== ===================
> 01.01.1900 15:11:10 01.01.1900 15:40:53
> 
> Strange behavior of the timezone
> 
> 
> 
> 
> select version()
> 
> version
> ========================================================================================================
> 
> PostgreSQL 9.6.5 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623
> (Red Hat 4.8.5-11), 64-bit
> 
> Thanks
> 
>

Re: Strange behavior of the timezone

От
Vlad Alexeenkov
Дата:
On server - UTC
Local - Europe/Moscow

For example, from 1920 all right:

If run in DataGrip:

select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 
15:11:10'::timestamp at time zone 'Europe/Moscow' as t2

1900-01-01 15:11:10.000000    1900-01-01 12:40:53.000000

select '1920-01-01 15:11:10'::timestamp t1, '1920-01-01 
15:11:10'::timestamp at time zone 'Europe/Moscow' as t2

1920-01-01 15:11:10.000000    1920-01-01 12:11:10.000000

select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 
15:11:10'::timestamp at time zone 'UTC-3' as t2

1900-01-01 15:11:10.000000    1900-01-01 12:11:10.000000

Thanks

On 27.03.2018 12:46, Robert Zenz wrote:
> That is not necessarily a strange behavior, as the Timezome most likely includes
> all the adjustments that were made in the period of time (leap seconds and the
> like). At least as far as I know.
>
> How does it behave if you try a more current date, like today? What is your
> system/database timezone?
>
>
> On 27.03.2018 11:40, Vlad Alexeenkov wrote:
>> Hi!
>>
>> Can anyone explain this behavior?
>>
>> select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 15:11:10'::timestamp at
>> time zone 'Europe/Moscow' as t2
>>
>>   t1                                       t2
>> =================== ===================
>> 01.01.1900 15:11:10 01.01.1900 15:40:53
>>
>> Strange behavior of the timezone
>>
>>
>>
>>
>> select version()
>>
>> version
>> ========================================================================================================
>>
>> PostgreSQL 9.6.5 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623
>> (Red Hat 4.8.5-11), 64-bit
>>
>> Thanks
>>



Re: Strange behavior of the timezone

От
Robert Zenz
Дата:
Here you go:

    https://www.timeanddate.com/time/change/russia/moscow?year=1919

That was what I was talking about, on 1st July, 1919 the clocks were turned
backwards by 31 minutes and 19 seconds. That accounts for the difference you see
here (plus or minus a leapsecond or two, or so, don't know). The remaining hours
difference are something about DST, but I'm too busy to look that up in detail.

What I wanted to say is that timezones do not only take *current* changes into
account, but also *past* changes. Actually, any known clock change is taken into
account, that includes leap seconds and any other adjustment of the clock (may
it be adding or removing time).


On 27.03.2018 13:04, Vlad Alexeenkov wrote:
> On server - UTC
> Local - Europe/Moscow
> 
> For example, from 1920 all right:
> 
> If run in DataGrip:
> 
> select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 15:11:10'::timestamp at
> time zone 'Europe/Moscow' as t2
> 
> 1900-01-01 15:11:10.000000    1900-01-01 12:40:53.000000
> 
> select '1920-01-01 15:11:10'::timestamp t1, '1920-01-01 15:11:10'::timestamp at
> time zone 'Europe/Moscow' as t2
> 
> 1920-01-01 15:11:10.000000    1920-01-01 12:11:10.000000
> 
> select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 15:11:10'::timestamp at
> time zone 'UTC-3' as t2
> 
> 1900-01-01 15:11:10.000000    1900-01-01 12:11:10.000000
> 
> Thanks
> 
> On 27.03.2018 12:46, Robert Zenz wrote:
>> That is not necessarily a strange behavior, as the Timezome most likely includes
>> all the adjustments that were made in the period of time (leap seconds and the
>> like). At least as far as I know.
>>
>> How does it behave if you try a more current date, like today? What is your
>> system/database timezone?
>>
>>
>> On 27.03.2018 11:40, Vlad Alexeenkov wrote:
>>> Hi!
>>>
>>> Can anyone explain this behavior?
>>>
>>> select '1900-01-01 15:11:10'::timestamp t1, '1900-01-01 15:11:10'::timestamp at
>>> time zone 'Europe/Moscow' as t2
>>>
>>>   t1                                       t2
>>> =================== ===================
>>> 01.01.1900 15:11:10 01.01.1900 15:40:53
>>>
>>> Strange behavior of the timezone
>>>
>>>
>>>
>>>
>>> select version()
>>>
>>> version
>>> ========================================================================================================
>>>
>>>
>>> PostgreSQL 9.6.5 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 4.8.5 20150623
>>> (Red Hat 4.8.5-11), 64-bit
>>>
>>> Thanks
>>>
> 
>

Re: autovacuum: vacuum analyze vs autovacuum: vacuum

От
Keith
Дата:


On Tue, Mar 27, 2018 at 5:04 AM, Artem Tomyuk <admin@leboutique.com> wrote:
Hi, Gurus!


As i understand from documentation:

when the autovacuum: vacuum analyze will be triggered, it will delete dead tuples and than will analyze table
but
when the autovacuum: vacuum will be triggered it will only mark for deletion dead tubles.

Is it valid explanation for postgres 9.4?

Thanks


I believe that when the ANALYZE portion of autovacuum kicks in, it only does the analyze. Which is why it's frequently tuned to run more often than the VACUUM portion of autovacuum since it's less of a load.

Keith

Re: autovacuum: vacuum analyze vs autovacuum: vacuum

От
Artem Tomyuk
Дата:
Ok, thanks for the clarification. 

вт, 27 бер. 2018, 18:48 користувач Keith <keith@keithf4.com> пише:


On Tue, Mar 27, 2018 at 5:04 AM, Artem Tomyuk <admin@leboutique.com> wrote:
Hi, Gurus!


As i understand from documentation:

when the autovacuum: vacuum analyze will be triggered, it will delete dead tuples and than will analyze table
but
when the autovacuum: vacuum will be triggered it will only mark for deletion dead tubles.

Is it valid explanation for postgres 9.4?

Thanks


I believe that when the ANALYZE portion of autovacuum kicks in, it only does the analyze. Which is why it's frequently tuned to run more often than the VACUUM portion of autovacuum since it's less of a load.

Keith