|
We have software which uses OpenStreetMap as a map and Nominatim as reverse geocoding to find exact street address for GPS point. A while ago we noticed that there had been major problem with one Nominatim server (pummelzacken) and it appeared so that our software was not able to connect to address nominatim.openstreetmap.org with proper parameters. The address redirected our query always to pummelzacken and connection attempt failed every time. Is there a reason why the query was redirected to pummelzacken when it was out of service? I see that server dulcy was up and running at same time. |
|
The planet database was reimported on both OSM Nominatim servers during that time, one after the other, which takes about 2 days each, so the redirect was temporary. Full reimports are rare, I think the last one was over 12 months ago. In preparation for Nominatim 3.6 release https://github.com/osm-search/Nominatim/releases/tag/v3.6.0 Thanks for the explanation. But if I have understood correctly, when server pummelzacken was under maintenance (the import process) other server dulcy should have been accessible in url https://nominatim.openstreetmap.org. Am I correct? At least uptime graph shows that dulcy was up between 27th Nov - 1st Dec when pummelzacken was under maintenance. Then I see dulcy was under maintenance between 6th Dec - 8th Dec and pummelzacken was up at that time. And the redirect. The address was redirected always to pummelzacken when pummelzacken was under maintenance. Do you confirm that? Do you report about the planet database reimport and other planned maintenance couple of days before the planned maintenance time? And also unplanned as soon as possible? For example in twitter.
(15 Dec '20, 06:53)
Pocum
1
I can't confirm, I dont' work in the OSM operations group. https://operations.osmfoundation.org/policies/nominatim/ has no uptime guarantees.
(15 Dec '20, 12:55)
mtmail
|
Just in case you haven't seen it: https://operations.osmfoundation.org/policies/nominatim/ .
No replies for a while but I would appreciate if someone can tell me if there has been any reason why the traffic was not redirected to dulcy server while pummelzacken was under maintenance brake (and vice versa). Of course any explanation is appreciated but I just want to know. I think that the problem is not in our end because it seems that the connection string is not cached for later use.