[Tile-serving] [openstreetmap/osm2pgsql] Tile Expiry Performance (#946)

[intergalacticdata]

> However, @intergalacticdata's statement "While expiring tiles, [the memory usage] ballooned to >50GB before being killed." is similarly unhelpful. Newer versions of osm2pgsql always balloon like this because of the memory-mapped flatnode file. It is not an indicator of a memory leak, nor reason for being OOM-killed.

In that instance, the process was manually interrupted rather than OOM killed.

I wonder if this is two different issues. My earliest tests against CVS builds (late May) failed with bad alloc() while later tests against 0.96 stayed in the range of 1GB resident memory if not expiring tiles. I did not test the CVS build without tile expiration.

Nakaner's recent tests are against a June build and take ~50GB whether tiles are expired or not.

@Nakaner, is there any chance you run one more test with a 0.96 release build and no tile expiration?

Unfortunately, I no longer have the capability to run these tests. The 50x increase in required RAM really seems broken though.

-- 
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/946#issuecomment-522683517
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20190819/3fe572d3/attachment.html>