stable TRAPI?

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

stable TRAPI?

Andre Hinrichs
Hi list!

Question to the developers of the TRAPI servers: When will the TRAPI
servers be stable again? The everlasting up and down of the servers is
very annoying.

Is this because of heavy testing or is the server unstable?

The rendering of prio 2 requests is now 17 hours behind. And
unfortunately this also means, that prio 3 and prio 4 requests are never
rendered.

And what about the server of flohoff? Is it permanentely down or will it
be available again in near future?


Regards
Andre



_______________________________________________
Tilesathome mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/tilesathome
Reply | Threaded
Open this post in threaded view
|

Re: stable TRAPI?

Florian Lohoff-2
On Mon, Feb 21, 2011 at 05:19:03PM +0100, Andre Hinrichs wrote:

> Hi list!
>
> Question to the developers of the TRAPI servers: When will the TRAPI
> servers be stable again? The everlasting up and down of the servers is
> very annoying.
>
> Is this because of heavy testing or is the server unstable?
>
> The rendering of prio 2 requests is now 17 hours behind. And
> unfortunately this also means, that prio 3 and prio 4 requests are never
> rendered.
>
> And what about the server of flohoff? Is it permanentely down or will it
> be available again in near future?
It is down and will likely be down permanently. I lost the possibility
of easy hosting as my employee was shutting the facility and layed
off all staff.

Flo
--
Florian Lohoff                                                 [hidden email]
                 Professionell gesehen bin ich zu haben ....

_______________________________________________
Tilesathome mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/tilesathome

signature.asc (845 bytes) Download Attachment
Reply | Threaded
Open this post in threaded view
|

Re: stable TRAPI?

Mathieu Arnold-2
In reply to this post by Andre Hinrichs
+--On 21 février 2011 17:19:03 +0100 Andre Hinrichs
<[hidden email]> wrote:
| Hi list!
|
| Question to the developers of the TRAPI servers: When will the TRAPI
| servers be stable again? The everlasting up and down of the servers is
| very annoying.
|
| Is this because of heavy testing or is the server unstable?

Well, I can only tell about my server, it's on it's knees, mainly because
TRAPI was designed when the database was way smaller, and I think TRAPI
can't cope well with the number of nodes we have, or the size of certain
z14 tiles.
Now, my server has another issue, that is it's quite full right now :

Filesystem  1K-blocks     Used    Avail Capacity iused ifree %iused
Mounted on
/dev/da0s1f  60812162 41013582 14933608    73% 7787820 78546   99%   /usr

--
Mathieu Arnold

_______________________________________________
Tilesathome mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/tilesathome
Reply | Threaded
Open this post in threaded view
|

Re: stable TRAPI?

Jeremy Adams
In reply to this post by Andre Hinrichs
On Mon, Feb 21, 2011 at 11:19 AM, Andre Hinrichs <[hidden email]> wrote:

> Hi list!
>
> Question to the developers of the TRAPI servers: When will the TRAPI
> servers be stable again? The everlasting up and down of the servers is
> very annoying.
>
> Is this because of heavy testing or is the server unstable?
>
> The rendering of prio 2 requests is now 17 hours behind. And
> unfortunately this also means, that prio 3 and prio 4 requests are never
> rendered.
>
> And what about the server of flohoff? Is it permanentely down or will it
> be available again in near future?
>
>
> Regards
> Andre

I'm not sure if there's an import or some mass editing going on right
now, but the minutely diffs are running between 100 and 200KB of
gzipped data.  I've noticed that my server is having trouble keeping
up with that rate of data.  It simply takes more than a minute to
import that much data and so it falls behind.  This is a dual CPU 3Ghz
Xeon w/ 8GB RAM and 4x 15K SAS drives.  I'm not sure if the solution
is to throw more hardware at it, or if there are tuning issues with
the TRAPI code since the database has grown so much since it was
written.

-Jeremy (milenko)

_______________________________________________
Tilesathome mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/tilesathome
Reply | Threaded
Open this post in threaded view
|

Re: stable TRAPI?

Daniel Bossert
 Hi list

> I'm not sure if there's an import or some mass editing going on right
> now, but the minutely diffs are running between 100 and 200KB of
> gzipped data.

> I'm not sure if the solution is to throw more hardware at it,
> or if there are tuning issues with the TRAPI code

> since the database has grown so much since it was written.

 So..I'm not a newbie but have not very long been active with OSM.

 My main parts with OSM is
 - having a t@h-client
 - mapping
 - fixing errors with keepright.imax.at


 Well, you say the database has grown. How about with optimizing the
 database?
 Look at this:
 http://www.openstreetmap.org/?lat=40.3320980072021&lon=16.7644929885864&zoom=14

 Way 87854044 (railway)
 A straightaway line with wasting 19 points. 2 points would do the same
 for this single straight line.
 This is for me a deal why the database grows. There are too many
 useless points..


 Are there some tools which could detect this?
 if there were a tool like keepright but for this type of erros we could
 optimize the database by deleting useless points.

 What do you think about my idea?


 I'm not a programmer and I don't have access to big servers. So I can't
 develop t@h further.

 But I would be willing to correct such types of errors (I'm already
 doing this when I see them)..


 Kind regards
 Daniel

_______________________________________________
Tilesathome mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/tilesathome
Reply | Threaded
Open this post in threaded view
|

Re: stable TRAPI?

dieterdreist
2011/2/22 Daniel Bossert <[hidden email]>:
> Look at this:
> http://www.openstreetmap.org/?lat=40.3320980072021&lon=16.7644929885864&zoom=14
>
> Way 87854044 (railway)
> A straightaway line with wasting 19 points. 2 points would do the same for
> this single straight line.


Well, for the current detail that might be true. When all bridges and
tunnels are mapped, all changing surfaces on minor roads, etc. you
will have much more nodes than in this link.

IMHO it is pointless to post a zoom 14 link and say that 19 points are
too much for a linear feature of about 5 km in the link (where nothing
else is mapped to give context). I agree that sometimes linear
features have too many nodes (especially most mappers don't use them
well, they tend to use regular distances regardless of curves or
straight parts) but I hope that if someone creates a tool to simplify
the geometry it will use angles and not just distances in its
algorithm. The current "simplify way"-features are deforming well
mapped ways.

In reality I see mostly the opposite: ways with far too few points (in
low version numbers). In respect to when all these ways will be mapped
with a more appropriate node density (and all the missing roads will
be present, that we are now lacking, e.g. in Asia, Australia, Africa)
deleting some pointless nodes is not a solution (e.g. see here:
http://www.openstreetmap.org/?lat=41.93815&lon=13.70882&zoom=17&layers=M
).

Of course I do delete pointless nodes, but that's far less than I add ;-)

cheers,
Martin

_______________________________________________
Tilesathome mailing list
[hidden email]
http://lists.openstreetmap.org/listinfo/tilesathome