Found some advice from Massimo in another thread.
After adding
try:
_junk = db
except NameError:
_w2pdir = "/Users/mellis/web2py/"
sys.path.append(_w2pdir)
from gluon import DAL
db = DAL("sqlite://storage.sqlite",
folder = _w2pdir + "applications/welcome/databases",
auto_import = True)
to my test script, running it without web2py gets the memory footprint
down from 25 Mb to 13Mb. That's definitely an improvement. Anyone
know how to shave it further? It still seems a little high for what
it's doing.
On Jan 2, 3:01 pm, Michael Ellis <[email protected]> wrote:
> Thanks Ross, that sounds like a good approach. I do something
> analogous in my app. All my scripts live in the app/modules/
> directory. My app is a remote monitoring system running on a Foxconn
> NetTop under Linux Mint 10. There is a web2py http interface used
> for installation and configuration, but it's not used in day-to-day
> operation.
> The long-running scripts are children of a master script similar to
> the one I appended to my post except that it has additional code to
> fork the various child processes which, conveniently, inherit the
> web2py environment including the model. The child processes write
> incoming data to the sqlitedb, read from it to compute averaged data
> that gets sent a server in the cloud, monitor for anomalous
> conditions and send email alerts, etc. The master script does little
> more than launch the children, monitor them, and relaunch any that
> crash.
> What arguments do you use in your calls to DAL() to read in your
> model? I just pass 'storage.sqlite' and DAL() knows which app/database
> directory to use since the app name is passed in the web2py -S
> invocation. Do you have to do anything special when importing dal to
> make it work without the web2py shell?
> Cheers,Mike
>
> On Jan 2, 1:32 pm, Ross Peoples <[email protected]> wrote:
>
>
>
>
>
>
>
> > Mike,
>
> > I haven't done any profiling on it, but I've had a lot of success using the
> > multiprocessing library, importing the DAL, and using it directly without
> > using the rest of the web2py environment. I usually accomplish this by
> > creating a module (not model) and spinning off the other processes from the
> > module, as the module will stay running after the initial request finishes.
>
> > Ross
