Il plante souvent, car on utilise l'instance publique sans garantie, qui semble avoir des gros problèmes de matos. https://github.com/valhalla/valhalla/issues/6067
Voilà les options :
Je pense que 3) est la voie à suivre, c'est la seule qui est pérenne pour nous.
Hi ! Is there a report of how much time it takes to prepare OSR for the whole planet on a standard 50 €/month VM ?
I'm asking because the Valhalla demo server has been down for 2 weeks and the dev has a hard time making it work again. So I thought, why not use OSR car mode without any public transport as a backup ?
the files can be downloaded from here (daily up-to-date):
https://api.transitous.org/gtfs/data/osr
Then it's mostly operating the server where I would guess that a 32GB instance is fine for planet scale routing.
Donc solution simple : implémenter OSR via Transitous puis télécharger les tuiles de Transitous pour l'opérer sur notre Motis qui tourne déjà !
À propos de la compilation de Valhalla chez nous :
and damn, you're building the planet in 4.5 hours? locally on mac or remote on a beast machine?
m8gd.4xlarge EC2 instance with 16 CPUs and 64GB RAM just for 1.1104$/h in eu-central-1 and 8gd instance family is important here because of really fast NVMe.
Our typical build takes a bit longer because of graph filtering, elevation, historical speeds - 6.5h in total.
I don't speak AWS, but sounds like a beast!
https://github.com/valhalla/valhalla/issues/6067#issuecomment-4552557526
Notre serveur 2 sur Scaleway c'est une bête aussi :), probablement nettement plus puissante.
No due date set.
No dependencies set.
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?