Yeah, we're not too likely to build in some elaborate support for such a niche use-case when there are tons of really universally useful things to work on. Let us know how it goes and if you run into any big problems.
Regards, Peter On Fri, Dec 10, 2010 at 12:35 PM, Zach Bailey <[email protected]> wrote: > Thank you David and Peter for your awesome ideas. Reading back over the > postgres COPY command docs [1] it doesn't look like that would ever be a > feasible solution given the following stipulation: > > *Files named in a **COPY** command are read or written directly by the > server, not by the client application. Therefore, they must reside on or be > accessible to the database server machine, not the client. They must be > accessible to and readable or writable by the **PostgreSQL** user (the > user ID the server runs as), not the client. **COPY** naming a file is > only allowed to database superusers, since it allows reading or writing any > file that the server has privileges to access.* > > So, it looks like this is something Heroku would have to wrap into an > abstracted administrative function down the line, if they ever did it at > all, due to the super-user access requirement. > > Given that, the remaining options consist of a.) restoring a partial backup > or b.) doing raw inserts. > > a.) seems like it would be possible using a dedicated db + heroku > pg:ingress + pg_restore -a -t tablename > > b.) is of course possible via a variety of methods (rake task, REST API, > taps, etc) but unfortunately rather slow as there is a lot of overhead in > doing single inserts (when compared to writing table structure directly, > etc.) > > I'm thinking I'll give option (a) a try and see how it goes. > > -Zach > > [1] http://www.postgresql.org/docs/8.4/interactive/sql-copy.html > > On Friday, December 10, 2010 at 3:12 PM, David Dollar wrote: > > Another possible solution would be this: > > Upload your data in CSV/TSV/whatever form to S3. Write a rake task that > does the following: > > * download from S3 to RAILS_ROOT/tmp > * use the psql command line tool (it's on our dyno grid) or one of the > ActiveRecord bulk import extensions to read the file and import to your > database > > Then you can run it with "heroku rake my_import_task" > > If this is going to be a regular process, you'll likely want to wrap all of > this up as something you can run from a worker using DJ or its' ilk. > > On Dec 10, 2010, at 10:51 AM, Zach Bailey wrote: > > > Thanks John, that's a great suggestion. Unfortunately it's looking like it > will take about 7.5 hours to import 3.12M rows: > > 1 tables, 3,123,800 records > companies: 1% | | ETA: > 07:25:34 > > I'm wondering if there's a more expedient route... in the past I've used > the postgres COPY command [1] to do bulk imports of large data sets quickly, > but that requires that the server be able to read a file off the server's > local filesystem. I don't suppose that's feasible given how the Heroku > platform works, but would love to be pleasantly surprised :) > > Anyone from Heroku able to pipe up and offer any other possible > suggestions? Just to restate the problem, I have a single table with about > 3.12M records that I'm wanting to transfer from a local DB to my remote > Heroku DB without touching the other Heroku app data. It's ok if the table > gets blown away on the Heroku side as it has nothing in it (new model I just > added). > > Happy Friday, > Zach > > [1] http://www.postgresql.org/docs/8.4/interactive/sql-copy.html > > On Thursday, December 9, 2010 at 4:36 AM, johnb wrote: > > If it's just a single table and you have it in a db locally then db:push > --tables <tablename> would get it up to heroku - but this will replace the > contents of the remote table with the local table and not append to it. If > the application is live you could put it into maintenance mode, db:pull > --tables <tablename> append your rows to it and then push the table back and > put the app live... > > perhaps? > > John. > > -- > You received this message because you are subscribed to the Google Groups > "Heroku" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/heroku?hl=en. > > > > -- > You received this message because you are subscribed to the Google Groups > "Heroku" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/heroku?hl=en. > > > -- > You received this message because you are subscribed to the Google Groups > "Heroku" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]. > For more options, visit this group at > http://groups.google.com/group/heroku?hl=en. > > > -- > You received this message because you are subscribed to the Google Groups > "Heroku" group. > To post to this group, send email to [email protected]. > To unsubscribe from this group, send email to > [email protected]<heroku%[email protected]> > . > For more options, visit this group at > http://groups.google.com/group/heroku?hl=en. > -- You received this message because you are subscribed to the Google Groups "Heroku" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/heroku?hl=en.
