[Talk-us] Great Lakes missing after osmosis update (diffs installation)

Pilon, Michel (SSC/SPC) michel.pilon at canada.ca
Tue Oct 3 19:39:30 UTC 2017


Thanks for responding Kevin,

Actually I have the issue after daily updating the whole planet.

But effectively, I don’t have that issue when I use only Ontario province or all Canada with geofabric daily updates…

Only with the whole planet daily update…. (same issue with hourly and minutely updates).

I am using a 32-cores server with 62 GB RAM and 1.4TB SSD.

ALL: Am I the only one downloading the whole planet and trying to daily update it successfully?

Please share your experience if you update the whole planet daily, hourly or minutely…

Thanks,

Michel




From: kevin.b.kenny at gmail.com [mailto:kevin.b.kenny at gmail.com] On Behalf Of Kevin Kenny
Sent: October 3, 2017 13:49
To: Pilon, Michel (SSC/SPC) <michel.pilon at canada.ca>
Cc: talk-us at openstreetmap.org
Subject: Re: [Talk-us] Great Lakes missing after osmosis update (diffs installation)

On Tue, Oct 3, 2017 at 1:31 PM, Pilon, Michel (SSC/SPC) <michel.pilon at canada.ca<mailto:michel.pilon at canada.ca>> wrote:
Have you gotten a chance to validate the coastlines of the Great Lakes during the last weekend?
You’re the only one who responded to my “help” request….

Oops, sorry, I thought I'd sent a reply.
I found one self-intersection on Lake Michigan and fixed it. Aside from that, there didn't
seem to be any problems, and the Great Lakes render all right on my machine. (Even
before the fix, they were rendering all right.)
I'm not a Mapnik maintainer, nor a really sophisticated user, so I'm not certain what
might be amiss. Could it be that you had a failed import of diffs that broke something?
(In which case, there might be no alternative to rerunning osm2pgsql from a planet
export.)
Does your app need up-to-the-minute global data? I usually run with just North America
from geofabrik, and only use geofabrik's daily diffs. I started out trying to maintain
a mirror with minutely diffs, and it was kind of a nightmare. If anything went wrong
even momentarily, it struggled to catch up and eventually failed. This was with a
quadcore, 32 GB main memory, 0.5 TB SSD / 4 TB RAID0+1. It may have been
bandwidth limited, but I think it was more that you need a bigger cache than I
could give it. That's a big database to try to synchronize remotely.

Otherwise, I'm kind of out of ideas. Maybe others who hear this message can
help more.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-us/attachments/20171003/070ccf3c/attachment-0001.html>


More information about the Talk-us mailing list