[Tile-serving] [openstreetmap/osm2pgsql] Drop slim mode tables just after finishing processing pending ways an… (#612)

[Paul Norman]

I ran a test import with this PR, and the commit before it

Import flags were `--cache 34500 --flat-nodes nodes.bin --slim` with a 161212 planet. The machine had 64GB RAM, vm.swappiness=20, and very fast NVMe drives

```
                master          PR
time            15:54:26        15:48:21
processing time 17650s          16641s
pending ways    6803s           6822s
indexing poly   14238s          13028s
indexing ways   32795s          33414s
Max res size    36.90 GB        36.87
```

No big differences, but but Munin is more interesting. Master is to the left, PR to the right.

![image](https://cloud.githubusercontent.com/assets/1190866/21466605/793071a8-c985-11e6-8aa6-fb5479683a92.png)

Just for reference, DB size is below on the same time scale
![image](https://cloud.githubusercontent.com/assets/1190866/21466620/e38cf9a4-c985-11e6-9d42-9ce5589fc6c5.png)

I could retry with swap disabled, but there's not much indication of memory differences with this testing.

Because with this much ram the node cache is unused, it probably all went out to swap.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/pull/612#issuecomment-269079976
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20161224/e0e0174f/attachment.html>