python shutting down sloooooooowly/tuning dictionaries
Till Plewe
till at score.is.tsukuba.ac.jp
Wed Apr 21 12:38:01 EDT 2004
More information about the Python-list mailing list
Wed Apr 21 12:38:01 EDT 2004
- Previous message (by thread): python shutting down sloooooooowly/tuning dictionaries
- Next message (by thread): python shutting down sloooooooowly/tuning dictionaries
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Wed, Apr 21, 2004 at 06:12:15PM +0200, Brian Quinlan wrote: > >Since the entire database is too large I typically run the same > >program 10-100 times on smaller parts. The annoying bit is that the > >time for finishing one instance of a program before starting the next > >instance can take a lot of time. ... > > You might try disabling garbage collection using the gc module when you > your application is done. Explicitly closing any files bound to globals > might then be a good idea. > > Cheers, > Brian > Thanks for the suggestions but unfortunately I already tried disabling gc. It did not have any noticable effect (not that I was patient enough to time it) nor did using sys.exit(). Also, I am not using threads which excludes another possible explanation. To give an example of the actual time spans involved: 2-10 minutes for running the program and > 30 minutes for shutdown. During that time the amount of memory used by python does not seem to vary. A far as closing files is concerned, I am using cdb for writing read-only databases and I am using cdb.finish() (there is no explicit method for closing(); finish writes all key/value pairs to a temporary file and then moves that file to wherever you want it. But as far as I can see as soon as that file is written/has been moved no more I/O should take place. There is no data loss if I simply kill the python program. - Till
- Previous message (by thread): python shutting down sloooooooowly/tuning dictionaries
- Next message (by thread): python shutting down sloooooooowly/tuning dictionaries
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Python-list mailing list