[Tile-serving] [openstreetmap/osm2pgsql] Very slow node parsing on updates ? (#1045)

[Christian Quest]

Here are some test results, all done on the same machine (24 cores, nvme SSD) on a France extract with 1 daily diff:

osm2pgsql | PG | nodes | ways | rels
--|--|--|--|--
0.96 | 10 | 60s | 33s | 23s
1.2.1 | 10 | 11s | 4s | 6s
0.96 | 12 | **575s** | 43s | 32s
1.2.1 | 12 | **276s** | 4s | 6s

I've looked at postgres changelog, and maybe found something to look at.
PG12 introduced a new parameter (plan_cache_mode) which is said to "allow control over when generic plans are used for prepared statements".
Instead of the "auto" default, I set it to force_generic_plan, it was event slower, then to force_custom_plan and got the same result as auto/default.
So... it really seems linked and related to some internal change in PG 12 :(

-- 
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/issues/1045#issuecomment-570159182
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20200102/f15df247/attachment-0001.html>