Finally, all services seem to be back in working shape again on the new server.
I don’t really know what went wrong in the first place, or what action actually fixed it, and for now I don’t care and just think about better monitoring instead.
Now to the new server. It’s a bit faster than the previous one, and has a little mor RAM, too, but most important: it has enough SSD space for all current services.
That makes planet database imports, and especially updates, much faster, and so the update lag has been minimized. The old server did fall behind at least once per day and could only catch up during less busy times, the new one so far has never been behind for more than 20 minutes, 15 of which are due to the update interval anyway.
Map rendering has also become a bit faster, especially multi page map booklet profit from this. I’m still considering to set up two separate rendering queues for single and multi page jobs though.
On the downside there’s less space to store rendered results, for now it still should be good enough to hold results for about a month before purging them. And the job entries will still be kept, so re-rendering can be requested easily.
The other downside is that I still don’t have enough space to maintain “old” and OsmCarto v.4.x osm2pgsql schema imports side by side, so for now the service is stuck with the last v.3.x style. That’s not too much of a pain for now, but over time printed and online map default styles will drift more and more apart.