The multilingular tiles are back, e.g.:
http://cassini.toolserver.org/tile-browse/browse-en.html?zoom=2&lat=40&a... http://cassini.toolserver.org/tile-browse/browse-is.html?zoom=2&lat=40&a...
I kicked it a bit, which involved:
* Recompiling mapnik/mod_tile/renderd from svn HEAD & installing. Mapnik now has gdal support and more goodies * rm -rf of the old tiles
We don't have cmarqu's hikingmap up yet though. mod_tile catches it but renderd never gets the requests, we're trying to find out why.
Also the text for the countries (and presumably all text) is a bit too low, compare the English map above with the main OSM one:
http://www.openstreetmap.org/?zoom=2&lat=40&lon=0
How were we on updates to the DB anyway?
How were we on updates to the DB anyway?
replication diff import is catching up, currently at 2009-11-09T08:31:03Z, so only some hours missing.
You can see the current state by looking at /sql/diffs/state.txt - you'll see the serial number as well as the date of the diff currently beeing imported. The Cronjob runs every 5 minutes. It locks itsself from running multiple times. You can see it's log by looking at /sql/diffs/logs/run.log
Atm most tile requests time out, maybe we should increase the timeout a little? Waiting longer until the loading finishs is better than getting a white page, i think.
Peter
On Mon, Nov 9, 2009 at 10:00 PM, Peter Körner osm-lists@mazdermind.de wrote:
How were we on updates to the DB anyway?
replication diff import is catching up, currently at 2009-11-09T08:31:03Z, so only some hours missing.
Sweet. We're getting a bunch of lines like this in the apache error log:
[Mon Nov 09 22:22:25 2009] [error] [client 94.194.60.19] Planet timestamp file (/sql/mod_tile//planet-import-complete)
Should we be creating this file?
You can see the current state by looking at /sql/diffs/state.txt - you'll see the serial number as well as the date of the diff currently beeing imported. The Cronjob runs every 5 minutes. It locks itsself from running multiple times. You can see it's log by looking at /sql/diffs/logs/run.log
Awesome.
Atm most tile requests time out, maybe we should increase the timeout a little? Waiting longer until the loading finishs is better than getting a white page, i think.
It was 10 and I changed it to 60. Let's see how that goes.
It'll also try to render tiles up to load 30 now, but won't re-render past load 4 like before.
Ævar Arnfjörð Bjarmason schrieb:
On Mon, Nov 9, 2009 at 10:00 PM, Peter Körner osm-lists@mazdermind.de wrote:
How were we on updates to the DB anyway?
replication diff import is catching up, currently at 2009-11-09T08:31:03Z, so only some hours missing.
Sweet. We're getting a bunch of lines like this in the apache error log:
[Mon Nov 09 22:22:25 2009] [error] [client 94.194.60.19] Planet timestamp file (/sql/mod_tile//planet-import-complete)
Should we be creating this file?
Touching this file invalidates all tiles without removing them, keeping them as fallback solution. Unsure what these messages say. So yes, maybe we should touch this file..
Awesome.
Should we move the cron to /etc/cron.d or should I keep it in my personal crontab?
echo '*/5 * * * * root /sql/diffs/load-next' > \ /etc/cron.d/load-next-diff
Peter
On Mon, Nov 9, 2009 at 10:37 PM, Peter Körner osm-lists@mazdermind.de wrote:
By they way we now have munin graphs of the box:
http://cassini.toolserver.org/munin/toolserver.org/cassini.toolserver.org.ht...
Ævar Arnfjörð Bjarmason schrieb:
On Mon, Nov 9, 2009 at 10:00 PM, Peter Körner osm-lists@mazdermind.de wrote:
How were we on updates to the DB anyway?
replication diff import is catching up, currently at 2009-11-09T08:31:03Z, so only some hours missing.
Sweet. We're getting a bunch of lines like this in the apache error log:
[Mon Nov 09 22:22:25 2009] [error] [client 94.194.60.19] Planet timestamp file (/sql/mod_tile//planet-import-complete)
Should we be creating this file?
Touching this file invalidates all tiles without removing them, keeping them as fallback solution. Unsure what these messages say. So yes, maybe we should touch this file..
mmmaybe:)
Awesome.
Should we move the cron to /etc/cron.d or should I keep it in my personal crontab?
echo '*/5 * * * * root /sql/diffs/load-next' > \ /etc/cron.d/load-next-diff
Ok. Removed this from your personal crontab and installed it in /etc/cron.d/load-next-diff
2009/11/9 Ævar Arnfjörð Bjarmason avarab@gmail.com:
Also the text for the countries (and presumably all text) is a bit too low, compare the English map above with the main OSM one:
I can confirm that (with a different style and a different font), so it's something to do with Mapnik maybe.
Cheers Colin
On Tue, Nov 10, 2009 at 12:16 AM, Colin Marquardt cmarqu42@googlemail.com wrote:
2009/11/9 Ævar Arnfjörð Bjarmason avarab@gmail.com:
Also the text for the countries (and presumably all text) is a bit too low, compare the English map above with the main OSM one:
I can confirm that (with a different style and a different font), so it's something to do with Mapnik maybe.
Yes I did upgrade mapnik to svn HEAD so I may have gotten a few bugs along with that :)
Ævar Arnfjörð Bjarmason schrieb:
The multilingular tiles are back, e.g.:
http://cassini.toolserver.org/tile-browse/browse-en.html?zoom=2&lat=40&a... http://cassini.toolserver.org/tile-browse/browse-is.html?zoom=2&lat=40&a...
I kicked it a bit, which involved:
- Recompiling mapnik/mod_tile/renderd from svn HEAD & installing.
Mapnik now has gdal support and more goodies
- rm -rf of the old tiles
Can renderd run in more than one process? We have 8 cores on cassini and only 1 of it can render tiles at a time. From looking at current top output I don't think that postgres is a bottleneck but renderd is.
Peter
Can renderd run in more than one process? We have 8 cores on cassini and only 1 of it can render tiles at a time. From looking at current top output I don't think that postgres is a bottleneck but renderd is.
I'm trying to get a Level 4 tile for about an hour now: http://cassini.toolserver.org/tiles/osm-like/de/4/8/7.png As well as some other Level 4 tiles.
Some neighboring tiles http://cassini.toolserver.org/tiles/osm-like/de/4/7/7.png http://cassini.toolserver.org/tiles/osm-like/de/4/8/8.png and some other Level 4 tiles are loading good
From top it seems that renderd is stopping rendering as soon as I get my 404 back. This is not the way it should be, is it?
Peter
On Tue, Nov 10, 2009 at 9:41 AM, Peter Körner osm-lists@mazdermind.de wrote:
Ævar Arnfjörð Bjarmason schrieb:
The multilingular tiles are back, e.g.:
http://cassini.toolserver.org/tile-browse/browse-en.html?zoom=2&lat=40&a... http://cassini.toolserver.org/tile-browse/browse-is.html?zoom=2&lat=40&a...
I kicked it a bit, which involved:
* Recompiling mapnik/mod_tile/renderd from svn HEAD & installing. Mapnik now has gdal support and more goodies * rm -rf of the old tiles
Can renderd run in more than one process? We have 8 cores on cassini and only 1 of it can render tiles at a time. From looking at current top output I don't think that postgres is a bottleneck but renderd is.
Yes it looks like we're underusing the CPU on the box. Last night I changed the number of renderd threads to 8 from 4. I'm not sure what the threads are doing exactly but I thought each one was rendering and that the master process handed out things to render.
There's also a distributed renderd patch available, but that's only if you're using multiple machines: http://trac.openstreetmap.org/ticket/2005
Ævar Arnfjörð Bjarmason schrieb:
The multilingular tiles are back, e.g.:
And we got problems with the size of some fonts:
http://cassini.toolserver.org/tile-browse/browse-th.html?zoom=3&lat=32.0...
Peter