[Tilesathome] My quad-core tiles at home setup

[Ævar Arnfjörð Bjarmason]

On Sat, Aug 23, 2008 at 12:20 PM, Dirk-Lüder Kreie <osm-list at deelkar.net> wrote:
> You are aware that every 4 hours (currently) there runs a autorequester
> on the t at h server that requests every tile with a node edit (moved,
> tagged, deleted or added) so your script is somewhat redundant?

Yes I'm aware that my tool is a fork and it's easy to stab yourself in
the eye with it:) I use it in cases when I don't want to wait up to
four hours to see how the area I just mapped turned out under
osmaprenderer. But in a lot of cases I don't use it and do what you
suggest, just wait.

> For the entire Project it would make more sense to run those clients in
> loop mode, but I can understand if you keep doing it your way,
> especially if you are the only really active mapper in a quite large
> area, and want to ensure your area is rendered by a homogenous client
> structure.

I usually run one of them in loop mode constantly (any more and I find
my machine starts swapping and everything goes wrong).

> However I would not recommend this behaviour for many other t at h clients,
> especially in areas where there are many mappers active, where the
> central queue has definitive advantages with respect to both render
> speed and resource conservation (API and t at h serverside)

I agree, if everyone was constantly re-rendering their little bbox and
nothing else the system as a whole would suffer.