Hi,
the grass website and wiki are both temporarily down due to a full disk. Hope to have it back up ASAP,it should be a quick fix after we figure out what's to blame.
Hamish
Hi,
the grass website and wiki are both temporarily down due to a full disk. Hope to have it back up ASAP,it should be a quick fix after we figure out what's to blame.
Hamish
Hamish:
the grass website and wiki are both temporarily down due to a full disk. Hope to
have it back up ASAP,it should be a quick fix after we figure out what's to
blame.
ok, they are back up now.
continued on the grass-dev list.
Hamish
Hamish wrote:
the grass website and wiki are both temporarily down due to a full disk. Hope to
have it back up ASAP,it should be a quick fix after we figure out what's to
blame.ok, they are back up now.
what seems to have happened is the "grass_addons.sh" cron job that compiles the addons (or extracts strings for translating?) got stuck in a loop building the help page for /home/martinl/src/grass-addons/grass6/raster/r.rdfilter/r.rdfilter.py.
r.rdfilter.py.tmp.html was inserting its own description.html over and over in a recursive loop until it filled the disk. I think I remember seeing a similar .tmp.html loop more than a year ago for some other module, probably not a Grass6 python addon, but I can't recall the exact circumstance other than it happened on my local system.
I'm not sure why it happened this time, but we should either figure out what caused it or switch off the cron job before it happens all over again, probably tomorrow night. I manually killed the job that got stuck.
Does it really have to run every 15 minutes?
Ideally, future builds could be run on a dedicated build server away from anything important. I think there's already wide acknowledgment of the need for that in OSGeo SAC. Note there is also currently the adhoc VM for things that may be risky or low priority, but be kind to the disk i/o as that is shared among all the other VMs on the same host.
regards,
Hamish
Hi,
2014-04-25 7:09 GMT+02:00 Hamish <hamish_b@yahoo.com>:
Hamish wrote:
the grass website and wiki are both temporarily down due to a full disk. Hope to
have it back up ASAP,it should be a quick fix after we figure out what's to
blame.ok, they are back up now.
what seems to have happened is the "grass_addons.sh" cron job that compiles the addons (or extracts strings for translating?) got stuck in a loop building the help page for /home/martinl/src/grass-addons/grass6/raster/r.rdfilter/r.rdfilter.py.
no, it's generating modules.xml files used by g.extension [1].
I'm not sure why it happened this time, but we should either figure out what caused it or switch off the cron job before it happens all over again, probably tomorrow night. I manually killed the job that got stuck.
I will turn off the job.
Ideally, future builds could be run on a dedicated build server away from anything important. I think there's already wide acknowledgment of the need for that in OSGeo SAC. Note there is also currently the adhoc VM for things that may be risky or low priority, but be kind to the disk i/o as that is shared among all the other VMs on the same host.
OK, it was discuss in the past with result that such cronjob is OK.
Things are changing, I can move the job to our servers and propagate
to grass.osgeo.org just resultant xml files...
Martin
I'm not sure why it happened this time, but we should either figure out what caused it or switch off the cron job before it happens all over again, probably tomorrow night. I manually killed the job that got stuck.
Does it really have to run every 15 minutes?
please bear in mind that it doesn't mean that it compiles addons every
15min. It just check every 15min if there is a new commit in SVN, than
it compiles. In any case the job is turned off. Later I will
investigate what was wrong (the cronjob was running more than one year
without any problem). Sorry for inconvenience.
Martin
--
Martin Landa * http://geo.fsv.cvut.cz/gwiki/Landa