[Osmf-talk] Core OpenStreetMap software development and support

Darafei "Komяpa" Praliaskouski me at komzpa.net
Thu Dec 21 09:00:04 UTC 2017


Hi Foundation members,

Many thanks to Jon Burgess for starting work to fix issues with
OpenStreetMap database replication. I hope that there are more people in
OSMF who can help with it.

https://github.com/openstreetmap/operations/issues/154#issuecomment-352768452


Not all things to fix are hard, but even simple ones can be beneficial for
infrastructure. If you know what is a queue and a hashmap (or will learn by
googling right after reading this message), you can help fixing the
rendering queue. I even set a monetary bounty for getting that done:

http://www.openstreetmap.org/user/Kom%D1%8Fpa/diary/42853


пн, 4 дек. 2017 г. в 13:47, Darafei "Komяpa" Praliaskouski <me at komzpa.net>:

> Hi,
>
> I'd like to flag an issue with our core software development process.
>
> openstreetmap.org is a complex system of many different servers being
> currently maintained by operations working group. Server hardware are being
> regularly taken care of, even though OWG consists of volunteers.
>
> If we look at https://github.com/openstreetmap we can see that project
> consists of many parts from software point of view. Some of them are
> maintained (notably JOSM, iD and openstreetmap-carto), but that's not the
> case for the rest. Let's have a look at what happened since June:
>
> openstreetmap-website, the rails application powering most of osm.org -
>  81 commits by Tom Hughes, mostly being security and policy tweaks;
>  19 commits by Bryan Housel, mostly being iD editor version bump;
>  18 commits by Andy Allan, mostly being refactorings.
> 277 open issues, 34 non-merged pull requests.
>
> https://github.com/openstreetmap/openstreetmap-website/graphs/contributors?from=2017-06-01&to=2017-12-04&type=c
>
>
> cgimap, c++ application powering /map call -
> no commits.
>
> https://github.com/openstreetmap/cgimap/graphs/contributors?from=2017-06-01&to=2017-12-04&type=c
>
>
> Both projects have major performance issues known for a year, with a
> prototype fix:
> https://github.com/Komzpa/fastmap. There is also an issue with
> replication diffs being implemented on top of unsuitable postgres mechanism
> that may cause data loss in diffs and resurrection of old data:
> https://github.com/openstreetmap/operations/issues/154
>
> mod_tile/renderd, c toolchain powering our tile servers:
> no commits. 77 issues raised. 19 pull requests waiting to be merged.
> has an issue of under-utilizing rendering server capacity.
> There is big comment on what needs to be improved by apmon:
> https://github.com/openstreetmap/mod_tile/pull/152#issuecomment-348805403
>
> osm2pgsql, c(++) application powering our tile servers:
> 10 commits. 84 issues. 8 pull requests open.
>
> https://github.com/openstreetmap/osm2pgsql/graphs/contributors?from=2017-06-01&to=2017-12-04&type=c
>
>
> Lack of resources to fix core issues is sometimes leading to workarounds
> that prevent growth:
>
>  - issues with scalability of rendering cluster led to bans and blocks of
> different web sites using openstreetmap's tiles;
>
>  - issue with API database scalability was worked around as "limit
> changeset size to less objects" instead of fixing the underlying API
> slowness and timing out issue. Discussions just blocked after merge:
> https://github.com/openstreetmap/openstreetmap-website/pull/1259
>
> OSMF was created as an organization to support the OpenStreetMap project
> and not shape the community. I think that this is precisely where OSM needs
> support and ask OSMF board to provide a way for us all to get out of the
> situation.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20171221/351011d6/attachment.html>


More information about the osmf-talk mailing list