On 04/04/2008 05:43 PM, mark wrote:
stats with new settings are below..
These stats look good for me.
but even with this sometimes update queries take more than coupla
seconds sometimes...
checkpoint_completion_target = 0.8
It looks like that this is set too high for your workload. If y
On Thu, Apr 3, 2008 at 10:02 PM, Greg Smith <[EMAIL PROTECTED]> wrote:
> On Wed, 2 Apr 2008, mark wrote:
>
> > with no clients connected to the database when I try to shutdown the
> > database [to apply new settings], it says database cant be shutdown.. for a
> > long time both in smart and normal
On Wed, 2 Apr 2008, mark wrote:
with no clients connected to the database when I try to shutdown the
database [to apply new settings], it says database cant be shutdown..
for a long time both in smart and normal mode... then i had to go to
immediate mode to shut down.. but then when i start it
Leif B. Kristensen wrote:
On Wednesday 2. April 2008, Tomasz Ostrowski wrote:
Go for it. Even 64 (I like round numbers) would not be too much.
Geek test: Do you find the above statement odd?
Yes: 0, No: +10.
(Sorry for being massively off-topic :-))
I had the same thought. ;)
Colin
--
On Thu, Apr 3, 2008 at 9:18 AM, Leif B. Kristensen <[EMAIL PROTECTED]> wrote:
> On Wednesday 2. April 2008, Tomasz Ostrowski wrote:
>
> >Go for it. Even 64 (I like round numbers) would not be too much.
>
> Geek test: Do you find the above statement odd?
Sadly, no.
--
Dave Page
EnterpriseDB UK
On Wednesday 2. April 2008, Tomasz Ostrowski wrote:
>Go for it. Even 64 (I like round numbers) would not be too much.
Geek test: Do you find the above statement odd?
Yes: 0, No: +10.
(Sorry for being massively off-topic :-))
--
Leif Biberg Kristensen | Registered Linux User #338009
http://solu
On Wed, Apr 2, 2008 at 1:19 AM, Greg Smith <[EMAIL PROTECTED]> wrote:
> On Wed, 2 Apr 2008, mark wrote:
>
> this really clear! Thannks!!
> >
>
> This is the first time someone new to this has ever said that about
> checkpoint tuning, which is quite the victory for all of us who worked
> toward th
On Wed, 2 Apr 2008, Tomasz Ostrowski wrote:
Even 64 (I like round numbers) would not be too much. This
would make 1GB of data in WALs.
It's much worse than that. Assume the system starts a checkpoint after
checkpoint_segments worth of WAL files have been used. It may take that
long before
On Wed, 2 Apr 2008, mark wrote:
this really clear! Thannks!!
This is the first time someone new to this has ever said that about
checkpoint tuning, which is quite the victory for all of us who worked
toward the 8.3 functional and monitoring improvements in this area.
Please keep posting the
On 2008-04-02 09:30, mark wrote:
> Based on what Tomasz suggested a day ago, I had changed settings to
>checkpoint_segments = 16
>checkpoint_timeout = 20min
>checkpoint_completion_target = 0.8
> but i still do get statements that take over 2 or 3 seconds to execute
> someti
On Tue, Apr 1, 2008 at 5:31 PM, Greg Smith <[EMAIL PROTECTED]> wrote:
> On Tue, 1 Apr 2008, mark wrote:
>
> current settings all default
> > > #checkpoint_segments = 3
> > > #checkpoint_timeout = 5min
> > > #checkpoint_completion_target = 0.5
> > > #checkpoint_warning = 30s
> > >
> >
> > this is
On Tue, 1 Apr 2008, mark wrote:
current settings all default
#checkpoint_segments = 3
#checkpoint_timeout = 5min
#checkpoint_completion_target = 0.5
#checkpoint_warning = 30s
this is what I have on pg_stat_bgwriter ; how much should I increase
checkpoint_segment & checkpoint_completion_target
mark <[EMAIL PROTECTED]> writes:
> I set this on,
> log_checkpoints = on
> and started postgres using this command
> pg_ctl -D /mnt/work/database -l /mnt/work/logs/pgsql.log start
> nothing is getting logged in the logfile. should I turn on some other
> settings?
You're probably looking in the
On Tue, Apr 1, 2008 at 7:27 AM, Tom Lane <[EMAIL PROTECTED]> wrote:
> Tomasz Ostrowski <[EMAIL PROTECTED]> writes:
> > I'd also set
> > log_checkpoints=on
> > to get an idea how it behaves.
>
> Yeah, that's really the *first* thing to do. You need to determine
>
I set this on,
log_checkpoin
Tomasz Ostrowski <[EMAIL PROTECTED]> writes:
> I'd also set
> log_checkpoints=on
> to get an idea how it behaves.
Yeah, that's really the *first* thing to do. You need to determine
whether the episodes of slowness are correlated with checkpoints
or not; there's no point fooling with the che
On Tue, Apr 1, 2008 at 1:48 AM, Tomasz Ostrowski <[EMAIL PROTECTED]>
wrote:
> On 2008-04-01 09:44, mark wrote:
>
> > I already am running 8.3.1 [ i mentioned in subject].
>
> But I have no experience on anything with more than 1GB of RAM...
>
Should I reduce shared_buffers to less than 1GB? If th
On 2008-04-01 09:44, mark wrote:
> I already am running 8.3.1 [ i mentioned in subject].
Missed that, sorry.
I'd first try to set the following on a write-busy 8.3 server to smooth
checkpoint spikes:
checkpoint_segments = 16
checkpoint_timeout = 20min
checkpoint_completio
On Tue, Apr 1, 2008 at 12:44 AM, mark <[EMAIL PROTECTED]> wrote:
> On Mon, Mar 31, 2008 at 11:18 PM, Tomasz Ostrowski <[EMAIL PROTECTED]>
> wrote:
>
> > On 2008-03-31 21:16, mark wrote:
> >
> > > is the query I am running , and it takes over 10 seconds to complete
> > > this query...
> > > update
On Mon, Mar 31, 2008 at 11:18 PM, Tomasz Ostrowski <[EMAIL PROTECTED]>
wrote:
> On 2008-03-31 21:16, mark wrote:
>
> > is the query I am running , and it takes over 10 seconds to complete
> > this query...
> > update users set number_recieved=number_recieved+1 where
> > uid=738889333;
>
> Every ti
On 2008-03-31 21:16, mark wrote:
> is the query I am running , and it takes over 10 seconds to complete
> this query...
> update users set number_recieved=number_recieved+1 where
> uid=738889333;
Every time or only sometimes?
If it is sometimes then I think this query is waiting for a checkpoin
On 31/03/2008 21:11, mark wrote:
it says actual time is 0.161 seconds or milliseconds.. but the total run
time is 11 seconds.. any ideas why this discrepancy?
Well, I interpret the docs as implying that the difference between the
time quoted in the top line of EXPLAIN ANALYZE's output and the
On Mon, Mar 31, 2008 at 12:59 PM, Raymond O'Donnell <[EMAIL PROTECTED]> wrote:
> On 31/03/2008 20:51, mark wrote:
>
> > can you explain what the numbers mean in the EXPLAIN ANALYZE?
> > (cost=0.00..8.46 rows=1 width=1073) (actual time=0.094..0.161 rows=1
> > loops=1)
>
> As I understand it, "cost
On Mon, Mar 31, 2008 at 12:59 PM, Raymond O'Donnell <[EMAIL PROTECTED]> wrote:
> On 31/03/2008 20:51, mark wrote:
>
> > can you explain what the numbers mean in the EXPLAIN ANALYZE?
> > (cost=0.00..8.46 rows=1 width=1073) (actual time=0.094..0.161 rows=1
> > loops=1)
>
> It's worth reading throug
On 31/03/2008 20:51, mark wrote:
can you explain what the numbers mean in the EXPLAIN ANALYZE?
(cost=0.00..8.46 rows=1 width=1073) (actual time=0.094..0.161 rows=1
loops=1)
It's worth reading through the docs at that reference in my previous
email - it's well explained there.
As I underst
On Mon, Mar 31, 2008 at 12:48 PM, Raymond O'Donnell <[EMAIL PROTECTED]> wrote:
> On 31/03/2008 20:38, mark wrote:
>
> > EXPLAIN ANALYZE update users set number_recieved=number_recieved+1 where
> > uid=738889333;
> > QUERY PLAN
> >
> -
On 31/03/2008 20:38, mark wrote:
EXPLAIN ANALYZE update users set number_recieved=number_recieved+1 where
uid=738889333;
QUERY PLAN
---
On 31/03/2008 20:38, mark wrote:
I dont manually do vaccum..
It might be worth doing one and seeing if it makes a difference.
Ray.
---
Raymond O'Donnell, Director of Music, Galway Cathedral, Ireland
[EMAIL PROTECTED]
---
On Mon, Mar 31, 2008 at 12:23 PM, Raymond O'Donnell <[EMAIL PROTECTED]> wrote:
> On 31/03/2008 20:16, mark wrote:
> > is the query I am running , and it takes over 10 seconds to complete
> > this query...
> >
> >
> > update users set number_recieved=number_recieved+1 where uid=738889333;
> >
> > t
On 31/03/2008 20:16, mark wrote:
is the query I am running , and it takes over 10 seconds to complete
this query...
update users set number_recieved=number_recieved+1 where uid=738889333;
table has about 1.7 million rows.. i have an index on column uid and
also on number_received. .. this is
29 matches
Mail list logo
