[Tilecache] how to speed up tilecache seeding in Tilecache_seed.py

[Alireza Kashian]

Hi.
Thanks to everyone.
I found the problem. As you guys mentioned , number of tiles showing in
tilecache_Seed output is actually divided by Metasize..I was not noticing to
that..and my second problem was that, I had two tilecache.cfg ..one in /etc/
and one in /etc/tilecache/tilecache-2.01

so I removed the one which didn't have correct config. I also found that
giving BBOX to tilecache is important.

now things are smooth and my Metatile is like 11x11 ...so around each 2
seconds, it generates 121 tiles..
in my calculations in one hour, my quad core CPU can generate 217800 tiles
:)

thanks
Alireza




On Wed, Jun 3, 2009 at 8:34 PM, Guillaume Sueur
<no-reply at neogeo-online.net>wrote:

> Hi,
>
> Also, please note that even when using Metatiling tilecache_seed stille
> counts how many tiles it has to generate, without indicating how many base
> tiles it corresponds to. What you can see in the tilecache_seed ouuput is
> that the total amount of tiles to generate is drastically lower than without
> metatile.
> But be sure tilecache_seed takes metatiling in consideration, cause it
> *does* !
>
> regards
>
> Guillaume
>
> Alireza Kashian a écrit :
>
>> Hi fredj
>>
>> Yes. I have done exactly same as you say.
>>
>> But are you using tilecache_seed.py code to seed? if yes, look at the
>> output screen, you see that tiles are generated one by one..
>> in my case, each tile takes 1 second to generate and we have like 40
>> layers in our map file..so I know it needs high computing power to send each
>> request seperately to tilecache...I was hoping that tilecache_seed.py could
>> jump another 11 tiles and request like 1 - 12 - 23 - 34 - 45 ....
>> because when it calls for 1, it generate all 121 tiles around it...when it
>> calls 12, it generate all tiles around 12...and so on..
>> so it doesn not need to call for: 1-2,3,4,5,6,7,8..........
>>
>> thanks
>> Alireza
>> On Wed, Jun 3, 2009 at 7:29 PM, Frederic Junod <
>> frederic.junod at camptocamp.com <mailto:frederic.junod at camptocamp.com>>
>> wrote:
>>
>>    Hi,
>>
>>    The meta tiling system works for me with the trunk version [1].
>>    Make sure you have the:
>>     metaTile=true
>>     metaSize=11,11
>>    in your layers options.
>>    Meta tiling is off by default even if you have a metaSize.
>>
>>    Regards,
>>    fredj
>>
>>
>>    PS: lucky you; I have more that 9'000'000 tiles in my last level :-)
>>
>>    [1] http://svn.tilecache.org/trunk/tilecache
>>
>>    On Wed, Jun 3, 2009 at 3:58 PM, Alireza Kashian
>>    <alireza.kashian at gmail.com <mailto:alireza.kashian at gmail.com>> wrote:
>>     > Hi.
>>     > I would like to know how we can making tiles faster in
>>    tilecache_seed.py
>>     >
>>     > The issue is that tilecache_seed.py tries to make tiles ONE by ONE..
>>     > so for example, in our case, in last zoom levels we have 300,000
>>    tiles and
>>     > each tile is generate in 1 second because our Map File is very
>>    complex and
>>     > parsing and making queries to PostGIS takes time.
>>     >
>>     > Actually MetaTiling is a very good solution, but the problem is
>>    that I feel
>>     > Tilecache_seed.py does its job without attention to Meta Tiling..for
>>     > example, our MetaTile is 11x11...so 121 Tiles are geneated in
>>    like 3-4
>>     > seconds...but When I run Tilecache_seed.py . I see from the log
>>    that it
>>     > makes tile requests one by one which takes like 300,000 seconds
>>    for just one
>>     > of our zoom levels..
>>     > Our map file is large , it is like 400 Kbyte..so I feel how hard
>>    it is to
>>     > make queries 300,000 times for making 300,000 tiles :)
>>     >
>>     > How can we help Tilecache_seed.py to benefit from MetaTile config in
>>     > Tilecache.cfg.
>>     >
>>     > Thanks
>>     > Alireza
>>     >
>>     >
>>     > _______________________________________________
>>     > Tilecache mailing list
>>     > Tilecache at openlayers.org <mailto:Tilecache at openlayers.org>
>>     > http://openlayers.org/mailman/listinfo/tilecache
>>     >
>>     >
>>
>>
>>
>>    --
>>    Frédéric Junod
>>    Camptocamp SA
>>
>>
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Tilecache mailing list
>> Tilecache at openlayers.org
>> http://openlayers.org/mailman/listinfo/tilecache
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.osgeo.org/pipermail/tilecache/attachments/20090604/bb9d912a/attachment.html