Hmm, looks like that should work automatically if you're using the Heroku contrib module to connect to your db. Are you doing:
from gluon.contrib.heroku import get_db db = get_db(name=None, pool_size=10) Anthony On Tuesday, September 1, 2015 at 11:48:27 PM UTC-4, James McGlynn wrote: > > Thanks for the response! > > I was able to find documentation on how web2py stores metadata in *.table > files. So I understand what you are saying about them, and I agree that > turning off migrations when running remotely would get rid of the error. > > But my question was about how/when web2py decides it needs to create the > "web2py_filesystem" table for storing metadata without a filesystem and > how/when it decides to look for metadata in that table, rather than the > file system. > > I found this answer > <https://groups.google.com/d/msg/web2py/TLlMMBYchGQ/SIdLQOPLTOYJ> from > Massimo on another thread: > > web2py stores metadata about DB tables which have been created. This is >> necessary for migrations. Normally this metadata is store in *.table files. >> Heroku has a strange filesystem that gets wiped up at random times. For >> this reason the *.table files are not in the "normal filesystem". Web2py >> creates a filesystem in database in a table called "web2py_filesystem" and >> stores the *.table files in there. > > > So I just would like to understand how this happens. Is there a place in > the source I could start reading that handles this? > > Thank you. > > > On Monday, August 31, 2015 at 9:48:59 PM UTC-4, Anthony wrote: >> >> The *.table files include metadata about the tables for use with the >> migrations functionality. If the *.table file is missing and migrations are >> turned on, web2py will assume the associated table needs to be created >> (resulting in an error if it already exists). Probably you simply want to >> turn off migrations in the production environment. >> >> Anthony >> >> On Monday, August 31, 2015 at 9:00:38 PM UTC-4, James McGlynn wrote: >>> >>> >>> class 'psycopg2.ProgrammingError' relation "my_table" already exists >>> >>> I have configured my app so that I can set a flag to use either a local >>> sqlite db, or a remote Heroku db when running locally, but when running >>> remotely on Heroku, the Heroku db is always used. The databases do not >>> contain the same data, nor do I need/want them to. >>> >>> I have come across this error a few times now for different reasons. I'm >>> pretty comfortable with resolving it at this point, but one way I >>> consistently get the error is by creating a new table while developing >>> locally with sqlite and then pushing my changes to Heroku, without first >>> running locally with the postgresql db. >>> >>> Can someone explain why my app seems to need a .table file for the new >>> table to be pushed to heroku, even though it gets eventually removed by >>> Heroku anyway? Does the logic for the storing metadata in the db use the >>> .table files in the filesystem to create the metadata stored in the db the >>> first time it senses that its running on Heroku or something? >>> >>> How would this work on a platform like elasticbeanstalk, which doesn't >>> offer the luxury of having the files on the file system temporarily >>> available after a push. >>> >> -- Resources: - http://web2py.com - http://web2py.com/book (Documentation) - http://github.com/web2py/web2py (Source code) - https://code.google.com/p/web2py/issues/list (Report Issues) --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to web2py+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
