[OSM-dev] dev server unresponsive

[Dirk-Lüder Kreie]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Nick Whitelegg schrieb:
> Doubt if this is the cause (as I have run it for about 8 weeks now without 
> problems) but I do run a couple of jobs on Wednesdays at 12 noon UK time:
> 
> - extract a UK-only planet.osm from planet.osm
> - run osm2pgsql on the UK-only planet.osm to generate PostGIS SQL (this is 
> used for Freemap)
> 
> Would these be better relocated to a quieter time of day or are these not 
> the problem?

- From my observations each use and cronjob by itself doesn't harm
operations, the problem seems to emerge when one job doesn't finish
before the other starts, and there is a higher than usual load from t at h
uploads.
The problem with t at h uploads is, that the clients will re-try uploading
until the upload succeeds, so if the server is already under load and
fails to cope with the increased upload load it will not recover,
because at some point the entire t at h crowd will be trying to upload,
because to the retries come the "normal" uploads. Of course there is a
limit to that because there is a limited number of t at h participants, but
apparently that was too much at least in this case.

- --

Dirk-Lüder "Deelkar" Kreie
Bremen - 53.0953°N 8.8651°E

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFGL2WPFUbODdpRVDwRAqcBAKCKv9saMiGTUGvGVPS4NGvZgx4VzwCeLJhH
5+Om7Kkecs4+/uDU0evXs7I=
=QcH4
-----END PGP SIGNATURE-----