So Webextra has a huge load right now, and I'm trying to figure out why.
A couple of observations:
1. None of the apache site configs write to their own log files, so it's
quite hard to figure out which sites are responsible for the load.
As a standard practice I recommend each site or collection of sites
writes to a separate log file (ie all the foss4g sites can go to one).
We should probably apply this idea to all of our servers.
2. The one monster apache log is full of
::1 - - [05/Dec/2012:07:47:19 -0800] "OPTIONS * HTTP/1.0" 200 - "-"
"Apache/2.2.9 (Debian) PHP/5.2.6-1+lenny16 with Suhosin-Patch (internal
dummy connection)"
Anyone know what these php internal dummy connections are and why their
might be so many?
Speaking of Foss4g.org sites on webextra, anyone know why there's a
section outside of Virtualhost tags?
Probably SAC should just rewrite the configs after waiting a few days?
I doubt that anybody of the site managers will react (sorry for being
pessimistic here)...
Probably SAC should just rewrite the configs after waiting a few days?
I doubt that anybody of the site managers will react (sorry for being
pessimistic here)...
Markus,
I think Alex has done so now. It looks like the main culprit on webextra
is the VMap Tile generation which was not documented in the wiki. I have
added it in:
It is using mapserver and some shapefiles to serve vmap0 tiles, presumably
mostly for openlayers, at the domain:
vmap0.tiles.osgeo.org
The traffic is variable, but it is not yet clear that it is causing serious
problems for the VM. I think we should consider:
1) Relaxing the munin load average alert.
2) See if we can limit tile service traffic in some way to avoid bogging
down the system serious if traffic continues to grow.
3) Consider finding a new home for this service (I think it already was
pushed here from the projects VM).
Best regards,
--
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam@pobox.com
light and sound - activate the windows | http://home.gdal.org/warmerda
and watch the world go round - Rush | Geospatial Software Developer
On Wed, Dec 05, 2012 at 12:46:21PM -0800, Frank Warmerdam wrote:
3) Consider finding a new home for this service (I think it already was
pushed here from the projects VM).
The "sphere.telascience.org" I use as MapServer for FlightGear Scenery
land cover has an adaption of this VMap0 tile service (the directory
structure is different) - but it's untested. If anyone would be
willing to select and describe a few reliable test cases, we might
consider moving it off the "webextra" VM or at least do some load
balancing.
Cheers,
Martin.
--
Unix _IS_ user friendly - it's just selective about who its friends are !
--------------------------------------------------------------------------
Probably SAC should just rewrite the configs after waiting a few days?
I doubt that anybody of the site managers will react (sorry for being
pessimistic here)...
Markus,
I think Alex has done so now. It looks like the main culprit on webextra
is the VMap Tile generation which was not documented in the wiki. I have
added it in:
On Wed, Dec 05, 2012 at 12:46:21PM -0800, Frank Warmerdam wrote:
I think Alex has done so now. It looks like the main culprit on webextra
is the VMap Tile generation which was not documented in the wiki. I have
added it in:
It is using mapserver and some shapefiles to serve vmap0 tiles, presumably
mostly for openlayers, at the domain:
vmap0.tiles.osgeo.org
I've fixed the setup on "sphere.telascience.org" and changed the DNS
accordingly.
Please test - and let's wait a few days to see how things develop.
Cheers,
Martin.
--
Unix _IS_ user friendly - it's just selective about who its friends are !
--------------------------------------------------------------------------