> Another idea, if you haven't tried it already, is to run these test cases
> in a server built with --enable-debug and --enable-cassert. The memory
> clobber stuff that's enabled by the latter is very good at turning coding
> errors into reproducible, debuggable crashes ;-)
>
>
Jeff Janes writes:
> On Fri, Jun 9, 2017 at 9:13 AM, Eric Lemoine
> wrote:
>> I now think that the performance bug is not related to the fn_extra
>> thing. I had hope but not anymore :) I don't see where the Pointcloud
>> and PostGIS extensions
On Fri, Jun 9, 2017 at 9:13 AM, Eric Lemoine
wrote:
> On 06/08/2017 10:41 PM, Éric wrote:
> >
> >
> >
> >> Have you experimented with other queries that don't involve PostGIS?
> >> I'm wondering if your hook-installation code fails to work properly
> >> unless PostGIS
On 06/09/2017 09:13 AM, Eric Lemoine wrote:
On 06/08/2017 10:41 PM, Éric wrote:
Have you experimented with other queries that don't involve PostGIS?
I'm wondering if your hook-installation code fails to work properly
unless PostGIS was loaded first. This would be easier to credit if
there
On 06/08/2017 10:41 PM, Éric wrote:
>
>
>
>> Have you experimented with other queries that don't involve PostGIS?
>> I'm wondering if your hook-installation code fails to work properly
>> unless PostGIS was loaded first. This would be easier to credit if
>> there are hooks both extensions try
>Have you experimented with other queries that don't involve PostGIS?
>I'm wondering if your hook-installation code fails to work properly
>unless PostGIS was loaded first. This would be easier to credit if
>there are hooks both extensions try to get into.
I think you're right on Tom. It
Eric Lemoine writes:
> The initial "select pc_typmod_pcid(1)" query completely screws the
> connection.
> "select pc_typmod_pcid(1)" is just an example of a simple query that
> triggers the problem. There are many others. But it has to be a query
> using the Pointcloud
On 06/08/2017 07:27 PM, Moreno Andreo wrote:
> Il 08/06/2017 19:10, Eric Lemoine ha scritto:
>>
>> How can such a thing happen? Thanks for any insight on what could cause
>> this.
>>
>>
> I'd try raising shared_buffers to 1 GB or something near 40% of the
> available memory
I tried to make it 4G,
Il 08/06/2017 19:10, Eric Lemoine ha scritto:
How can such a thing happen? Thanks for any insight on what could cause
this.
I'd try raising shared_buffers to 1 GB or something near 40% of the
available memory
If you run the query again, after getting bad results, what do you get?
Cheers
> Note that the execution time is 46 ms when the query is wrapped in an
> explain analyze (while it's 3 s when it's not!)
Actually, it seems to me that the performance issue is not on the query
itself, it is on the fetching of the data returned by the query. Which
explains why the query is fast
> Lots of missing information here ...
>
> Is there an index on public.sthelens.points?
Yes, there are.
lopocs=# \d sthelens;
Table "public.sthelens"
Column |Type| Modifiers
On 06/08/2017 09:20 AM, Eric Lemoine wrote:
Looks like you also have postgis and pointcloud_postgis in mix. I would
say this may get an answer sooner here:
http://lists.osgeo.org/mailman/listinfo/pgpointcloud/
I am actually one of the developers of the Pointcloud extension. I
haven't been
> Looks like you also have postgis and pointcloud_postgis in mix. I would
> say this may get an answer sooner here:
>
> http://lists.osgeo.org/mailman/listinfo/pgpointcloud/
I am actually one of the developers of the Pointcloud extension. I
haven't been able to debug this up to now.
--
Éric
On 06/08/2017 09:00 AM, Eric Lemoine wrote:
Hi
We have a rather strange performance issue with the Pointcloud extension
[*]. The issue/bug may be in the extension, but we don't know for sure
at this point. I'm writing to the list to hopefully get some guidance on
how to further debug this.
[*]
On Thu, 8 Jun 2017 18:00:04 +0200
Eric Lemoine wrote:
> We have a rather strange performance issue with the Pointcloud extension
> [*]. The issue/bug may be in the extension, but we don't know for sure
> at this point. I'm writing to the list to hopefully get some
15 matches
Mail list logo