[Tile-serving] [osm2pgsql] Very low performace on adding diff-s (#343)

[Paul Norman]

Processing node times are probably the easiest to look at, as it's one of the simplest stages. With `--number-processes 2 -v -G -K -j -x -s --append` it's going to be writing the node info to the nodes table and some of those into the points table.

`-x -j` is interesting, because I recall reports of that combination being very slow since it was writing full details of every node into the points table (-x adding tags, and -j putting them into hstore). But it wouldn't cause run to run variations.

None of this really helps identify the cause of the variations, and the log doesn't provide enough information.

Do you have logging (nagios, munin, etc) for the first two runs?



---
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/issues/343#issuecomment-96027569
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20150424/3999ef6b/attachment.html>