Sorry I misunderstood the problem. There is no way cron can rely on the existence of a web server running. yet there should be an option to do what you suggest (but the URL would have to be provided).
On Jul 28, 8:46 am, Iceberg <[email protected]> wrote: > That is not the case. Let me better clarify my purpose in pseudo code. > How about changing line 277~282 of gluon/newcron.py from: > > elif action: > commands.extend(('-P', > '-N',models, > '-S',app+'/'+command, > '-a','"<recycle>"')) > shell = True > > to: > > elif action: > urllib.urlopen('http://localhost:%s/%s/%s'% ( > server_port, # but how to know this value? > app, > command > ) ).read() > continue > > I don't think request.env.server_port is available in the newcron.py > context. Does it? > > Regards, > Iceberg > > On Jul 28, 8:46pm, mdipierro <[email protected]> wrote: > > > You can do what you suggest. You cant find the post in > > request.env.server_port > > You see all variables here : > > >http://web2py.com/examples/simple_examples/status > > > yet I feel there may be a better way using a queue and a background > > process (without CRON). It depends on what this is really for. > > > On Jul 28, 5:30 am, Iceberg <[email protected]> wrote: > > > > But cache.ram is theoretically faster than cache.disk. > > > > Besides, what is the pros and cons of using urllib.urlopen('http:// > > > localhost:8000/myapp/...').read() to visit an action? My draft: > > > > Pros: > > > - There is no need to app's environment for the cron job to run, hence > > > avoids potential locking problems radically. > > > - Just a urlopen(...).open() is considered less overhead than a full > > > "web2py -P -N -S ..." > > > > Cons: > > > - I did not find a way to auto detect which port the main web2py > > > process is running at. > > > > On Jul 28, 6:06pm, Massimo Di Pierro <[email protected]> wrote: > > > > > The solution is cache.disk > > > > > > PS: Using cache.disk() is not (yet) a workaround for > > > > web2py_win.zip. I > > > > > report a bug about that in another post. > > > > > I do not recall this. Can you tell us more and point to the post? > > > > > Massimo > > > > > On Jul 28, 2010, at 4:48 AM, Iceberg wrote: > > > > > > Scenario: In one of my small enterprise app, my action is time > > > > > consuming, so I use cache. Furthermore, I setup a cron job to "touch" > > > > > this action frequently, hoping that other real users will be benefited > > > > > by always have cache hits. Example as below. > > > > > > @cache(request.env.path_info,time_expire=1800,cache_model=cache.ram) > > > > > def index(): > > > > > result = do_a_time_consuming_job() > > > > > return {'result': result} > > > > > > # the crontab file > > > > > */30 * * * * root **default/index > > > > > > Nice plan, huh? But it has no effect. Because each time the cronjob is > > > > > seemingly invoked in a separated process space (via "web2py -P -N -M - > > > > > S myapp/default/index -a <recycled>"), so the main process's cache.ram > > > > > is not touched. > > > > > > So, is it possible to change the implementation, let the "visit my > > > > > action" cronjob be done by this way? > > > > > urllib.urlopen('http://localhost:8000/myapp/default/index').read() > > > > > > By the way, my current workaround is to write my crontab file like > > > > > this: > > > > > */30 * * * * root > > > > > **applications/myapp/cron/trigger.py > > > > > and my trigger.py is based on urlopen(...).read() mentioned above. > > > > > The workaround works, but I don't like the fact that it hardcode the > > > > > app name (as well as the port). > > > > > > PS: Using cache.disk() is not (yet) a workaround for web2py_win.zip. I > > > > > report a bug about that in another post. > > > > > > So, what's your opinion? > > > > > > Regards, > > > > > Iceberg > > > > > > -- > > > > > mail from:GoogleGroups "web2py-developers" mailing list > > > > > make speech: [email protected] > > > > > unsubscribe: [email protected] > > > > > details :http://groups.google.com/group/web2py-developers > > > > > the project:http://code.google.com/p/web2py/ > > > > > official :http://www.web2py.com/
