Обсуждение: Can we Flush the Postgres Shared Memory ?

On Tue, May 3, 2011 at 11:54 AM, raghu ram <raghuchennuru@gmail.com> wrote:

> It may be a silly question, still out of curiosity I want to know, is there
> any possible way to flush the Postgres Shared Memory without restarting the
> cluster.
> In Oracle, we can flush the SGA, can we get the same feature here..
> Thanks in Advance.


The CHECKPOINT command will do this for you.

--
 Simon Riggs                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services

On Tue, May 3, 2011 at 8:30 AM, raghu ram <raghuchennuru@gmail.com> wrote:
> On Tue, May 3, 2011 at 6:01 PM, Raghavendra
> <raghavendra.rao@enterprisedb.com> wrote:
>>
>> On Tue, May 3, 2011 at 5:37 PM, Simon Riggs <simon@2ndquadrant.com> wrote:
>>>
>>> On Tue, May 3, 2011 at 11:54 AM, raghu ram <raghuchennuru@gmail.com>
>>> wrote:
>>>
>>> > It may be a silly question, still out of curiosity I want to know, is
>>> > there
>>> > any possible way to flush the Postgres Shared Memory without restarting
>>> > the
>>> > cluster.
>>> > In Oracle, we can flush the SGA, can we get the same feature here..
>>> > Thanks in Advance.
>>>
>>>
>>> The CHECKPOINT command will do this for you.
>>
>>
>
> According to PostgreSQL documentation, whenever you execute "CHECKPOINT" in
> the database,it will flush the modified data files presented in the Shared
> Buffers retuned to the Disk.
>            http://www.postgresql.org/docs/9.0/static/sql-checkpoint.html
> Is this clears the entire shared memory cache and same time,if i execute
> fresh SQL statement, Data will be retuned from disk??

no it will not, or at least there is no guarantee it will be.  the
only way to reset the buffers in that sense is to restart the database
(and even then they might not be read from disk, because they could
sit in the o/s cache).  to force a read from the drive you'd have to
reboot the server, or at least shut it down and use a lot of memory
for some other purpose.

merlin

> On Tue, May 3, 2011 at 6:01 PM, Raghavendra <
> raghavendra.rao@enterprisedb.com> wrote:
>
>>
>> On Tue, May 3, 2011 at 5:37 PM, Simon Riggs <simon@2ndquadrant.com>
>> wrote:
>>
>>> On Tue, May 3, 2011 at 11:54 AM, raghu ram <raghuchennuru@gmail.com>
>>> wrote:
>>>
>>> > It may be a silly question, still out of curiosity I want to know, is
>>> there
>>> > any possible way to flush the Postgres Shared Memory without
>>> restarting
>>> the
>>> > cluster.
>>> > In Oracle, we can flush the SGA, can we get the same feature here..
>>> > Thanks in Advance.
>>>
>>>
>>> The CHECKPOINT command will do this for you.
>>>
>>
>>
>>
> According to PostgreSQL documentation, whenever you execute "CHECKPOINT"
> in
> the database,it will flush the modified data files presented in the Shared
> Buffers retuned to the Disk.
>
>            http://www.postgresql.org/docs/9.0/static/sql-checkpoint.html
>
> Is this clears the entire shared memory cache and same time,if i execute
> fresh SQL statement, Data will be retuned from disk??

No. Checkpoint means all dirty buffers are written to the datafiles, it
does not mean emptying the shared buffers. Checkpoints happen regularly so
this would have an unwanted impact on performance.

And besides that, there's a page cache maintained by the OS (not sure if
you're running Linux or Windows). So even when the block does not exist in
the shared buffers, it may be in the page cache (thus not read from the
drive).

Dropping the page cache is quite simple (http://linux-mm.org/Drop_Caches),
emptying the shared buffers is not that simple - I guess the easiest way
is to restart the db.

What are you trying to achieve? Why do you need this?

Tomas

2011/5/3 Merlin Moncure <mmoncure@gmail.com>:
> On Tue, May 3, 2011 at 8:30 AM, raghu ram <raghuchennuru@gmail.com> wrote:
>> On Tue, May 3, 2011 at 6:01 PM, Raghavendra
>> <raghavendra.rao@enterprisedb.com> wrote:
>>>
>>> On Tue, May 3, 2011 at 5:37 PM, Simon Riggs <simon@2ndquadrant.com> wrote:
>>>>
>>>> On Tue, May 3, 2011 at 11:54 AM, raghu ram <raghuchennuru@gmail.com>
>>>> wrote:
>>>>
>>>> > It may be a silly question, still out of curiosity I want to know, is
>>>> > there
>>>> > any possible way to flush the Postgres Shared Memory without restarting
>>>> > the
>>>> > cluster.
>>>> > In Oracle, we can flush the SGA, can we get the same feature here..
>>>> > Thanks in Advance.
>>>>
>>>>
>>>> The CHECKPOINT command will do this for you.
>>>
>>>
>>
>> According to PostgreSQL documentation, whenever you execute "CHECKPOINT" in
>> the database,it will flush the modified data files presented in the Shared
>> Buffers retuned to the Disk.
>>            http://www.postgresql.org/docs/9.0/static/sql-checkpoint.html
>> Is this clears the entire shared memory cache and same time,if i execute
>> fresh SQL statement, Data will be retuned from disk??
>
> no it will not, or at least there is no guarantee it will be.  the
> only way to reset the buffers in that sense is to restart the database
> (and even then they might not be read from disk, because they could
> sit in the o/s cache).  to force a read from the drive you'd have to
> reboot the server, or at least shut it down and use a lot of memory
> for some other purpose.

with linux, you can : "echo 3 > /proc/sys/vm/drop_caches" for the OS cache

>
> merlin
>
> --
> Sent via pgsql-admin mailing list (pgsql-admin@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-admin
>



--
Cédric Villemain               2ndQuadrant
http://2ndQuadrant.fr/     PostgreSQL : Expertise, Formation et Support

2011/5/3 Merlin Moncure <mmoncure@gmail.com>:
> On Tue, May 3, 2011 at 8:55 AM, Cédric Villemain
> <cedric.villemain.debian@gmail.com> wrote:
>> 2011/5/3 Merlin Moncure <mmoncure@gmail.com>:
>>>
>>> no it will not, or at least there is no guarantee it will be.  the
>>> only way to reset the buffers in that sense is to restart the database
>>> (and even then they might not be read from disk, because they could
>>> sit in the o/s cache).  to force a read from the drive you'd have to
>>> reboot the server, or at least shut it down and use a lot of memory
>>> for some other purpose.
>>
>> with linux, you can : "echo 3 > /proc/sys/vm/drop_caches" for the OS cache
>>
>
> yeah -- good point.  aside: does that also drop cache on the drive/raid card?

no -- good point too ! (damn! how SAN users will do...maybe EMC or
other are good enough to provide some control panel for that ? )

and as I read on the link provided by Tomas, it is better to issue a
'sync' before trying to drop cache (I do that sometime, but postgresql
flush its write before shutdown, so I expected the dirty pages in OS
cache not to be relative to postgresql files.)


--
Cédric Villemain               2ndQuadrant
http://2ndQuadrant.fr/     PostgreSQL : Expertise, Formation et Support

On Tue, May 3, 2011 at 2:30 PM, raghu ram <raghuchennuru@gmail.com> wrote:

>>> The CHECKPOINT command will do this for you.
>>
>>
>
> According to PostgreSQL documentation, whenever you execute "CHECKPOINT" in
> the database,it will flush the modified data files presented in the Shared
> Buffers retuned to the Disk.
>            http://www.postgresql.org/docs/9.0/static/sql-checkpoint.html
> Is this clears the entire shared memory cache and same time,if i execute
> fresh SQL statement, Data will be retuned from disk??

No, but then you'd need to flush OS buffers and all disk caches as
well to make that effective.

--
 Simon Riggs                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services