Re: [OSM-talk] High load on the rendering servers?
On Sun, 08 Mar 2015 15:45:29 +0100, colliar wrote: Sounds like work for the editor software if the api is not capable to prevent moving a node hundreds of kilometres. Even if the API rejects very long ways the editor needs a little work to output a useful error message like precondition failed (just love that one) to the user. Thomas ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] High load on the rendering servers?
On 08/03/15 04:28, Eugene Alvin Villar wrote: On Sun, Mar 8, 2015 at 6:15 AM, Simon Poole si...@poole.ch mailto:si...@poole.ch wrote: Jon Burgess tracked the issue down to a way node being dragged from Japan to Brazil on the 3rd of this month, creating a very very long way that increased the rendering time for a large number of tiles at high zoom levels. Wow. This constitutes an extremely simple way to mount a DoS attack on the OSM tileservers. I will have to assume that the Operations WG is already thinking of ideas on how to prevent this possibility in the future. I think I can safely say we're not. Of course if you have a brilliant idea how to do such a thing then I'm sure we'd love to hear it. Tom -- Tom Hughes (t...@compton.nu) http://compton.nu/ ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] High load on the rendering servers?
On Sunday 08 March 2015, Tom Hughes wrote: Wow. This constitutes an extremely simple way to mount a DoS attack on the OSM tileservers. I will have to assume that the Operations WG is already thinking of ideas on how to prevent this possibility in the future. I think I can safely say we're not. Of course if you have a brilliant idea how to do such a thing then I'm sure we'd love to hear it. The OSM inspector already has a display mode for ways with long segments - it currently however seems it is not working: http://tools.geofabrik.de/osmi/?view=geometryoverlays=ways_with_long_segments,long_segments And it has a fairly low threshold (0.3 degrees - that is ~40km) so it highlights quite a lot of stuff. With an additional layer highlighting only the very long segments (like above 300km segment length) you could quickly see such issues. And editors of course also should prominently warn about such edits. -- Christoph Hormann http://www.imagico.de/ ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] High load on the rendering servers?
Am 08.03.2015 um 10:20 schrieb Christoph Hormann: On Sunday 08 March 2015, Tom Hughes wrote: Wow. This constitutes an extremely simple way to mount a DoS attack on the OSM tileservers. I will have to assume that the Operations WG is already thinking of ideas on how to prevent this possibility in the future. I think I can safely say we're not. Of course if you have a brilliant idea how to do such a thing then I'm sure we'd love to hear it. The OSM inspector already has a display mode for ways with long segments - it currently however seems it is not working: http://tools.geofabrik.de/osmi/?view=geometryoverlays=ways_with_long_segments,long_segments And it has a fairly low threshold (0.3 degrees - that is ~40km) so it highlights quite a lot of stuff. With an additional layer highlighting only the very long segments (like above 300km segment length) you could quickly see such issues. And editors of course also should prominently warn about such edits. Sounds like work for the editor software if the api is not capable to prevent moving a node hundreds of kilometres. cu colliar [1] https://josm.openstreetmap.de/ticket/11215 0xE8F56581.asc Description: application/pgp-keys signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] High load on the rendering servers?
Hi Andrew, Yes were are aware there is an issue. We haven't yet tracked down the issue. We have been discussing it in #osm-dev on http://irc.openstreetmap.org today. Kind regards, Grant Part of the OSM sysadmin On 6 March 2015 at 15:20, Andrew Guertin andrew.guer...@uvm.edu wrote: For the past few days, lots of things I've changed haven't had their tiles re-rendered, and I noticed that the servers are reporting very high load and lots of dropped tiles: http://munin.openstreetmap.org/renderd-week.html Based on my (completely uneducated) reading of the graphs there, it looks like something is filling the Priority Request Queue and keeping it full, and there's very little time for anything else. (It looks like the Request Queue and the Low Priority Request Queue are also being kept full). Anyone know what's causing this? --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] High load on the rendering servers?
Jon Burgess tracked the issue down to a way node being dragged from Japan to Brazil on the 3rd of this month, creating a very very long way that increased the rendering time for a large number of tiles at high zoom levels. Simon Am 07.03.2015 um 21:39 schrieb Grant Slater: Hi Andrew, Yes were are aware there is an issue. We haven't yet tracked down the issue. We have been discussing it in #osm-dev on http://irc.openstreetmap.org today. Kind regards, Grant Part of the OSM sysadmin On 6 March 2015 at 15:20, Andrew Guertin andrew.guer...@uvm.edu wrote: For the past few days, lots of things I've changed haven't had their tiles re-rendered, and I noticed that the servers are reporting very high load and lots of dropped tiles: http://munin.openstreetmap.org/renderd-week.html Based on my (completely uneducated) reading of the graphs there, it looks like something is filling the Priority Request Queue and keeping it full, and there's very little time for anything else. (It looks like the Request Queue and the Low Priority Request Queue are also being kept full). Anyone know what's causing this? --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk signature.asc Description: OpenPGP digital signature ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
Re: [OSM-talk] High load on the rendering servers?
On Sun, Mar 8, 2015 at 6:15 AM, Simon Poole si...@poole.ch wrote: Jon Burgess tracked the issue down to a way node being dragged from Japan to Brazil on the 3rd of this month, creating a very very long way that increased the rendering time for a large number of tiles at high zoom levels. Wow. This constitutes an extremely simple way to mount a DoS attack on the OSM tileservers. I will have to assume that the Operations WG is already thinking of ideas on how to prevent this possibility in the future. ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk
[OSM-talk] High load on the rendering servers?
For the past few days, lots of things I've changed haven't had their tiles re-rendered, and I noticed that the servers are reporting very high load and lots of dropped tiles: http://munin.openstreetmap.org/renderd-week.html Based on my (completely uneducated) reading of the graphs there, it looks like something is filling the Priority Request Queue and keeping it full, and there's very little time for anything else. (It looks like the Request Queue and the Low Priority Request Queue are also being kept full). Anyone know what's causing this? --Andrew ___ talk mailing list talk@openstreetmap.org https://lists.openstreetmap.org/listinfo/talk