FAQ

We try to keep this Frequently Asked Questions page up-to-date based on feedback from people trying Checkbook, and we welcome submissions for new FAQ items.

• What's the best place to ask questions or make suggestions about Checkbook?

  The best ways to ask questions about Checkbook are:

  • File an issue here. We'll see it and respond (it helps if you mark it with the "question" label).
  • Post your question in the CheckbookNYC discussion forum at Google Groups. You can also post to the group by email: checkbooknyc {_AT_} googlegroups.com.
  • See the CheckbookNYC user help forum at support.mymoneynyc.com/mymoneynyc/products/mymoneynyc_checkbook_20. (This is what Help->FAQ leads to from the CheckbookNYC.com home page.)

• How do I report a bug in Checkbook?

  Right here, in the bug tracker, please. Remember to look first to see if your issue has already been filed by someone else.

• How hard is it to set up a production Checkbook instance? How much does it cost?

  We don't have enough experience with multiple different instances to give a good answer to this question yet (as of 2013-09), and naturally are reluctant to risk spreading any misinformation. However, please watch this space for updates, as we intend to give a better answer, including approximate dollar figures, as soon as we responsibly can.

• How much does it cost to just try Checkbook, such as in a sandbox instance?

  Well, Checkbook is open source, so trying it costs nothing, right?

  Just kidding -- sysadmin time costs money, as do servers, as does testing, as does every aspect of the process. Until issue #19 is resolved, it's hard for us to give a useful estimate, since every municipality's data import process may be different, either a little bit different or a lot different. However, cities besides New York are trying out Checkbook right now, and we're in contact with them, so we will update this item as we have more data, hopefully soon (as of 2013-09).

• What are the minimum hardware/bandwidth requirements for running Checkbook?

  The answer will depend on the size of your financial data and on the expected usage (peak request bandwidth, etc) of your production instance.

  Right now, the sizing of the Java heap for Solr is done by trial and error. For New York City's data (index size currently about 20G) the following parameters are used in /opt/apache-tomcat-6.0.35/bin/catalina.sh:

  -Xms7130M -Xmx7130M -XX:+UseConcMarkSweepGC -XX:-UseParNewGC

  We have filed issue #16 about better documenting server requirements. When we have a better handle on that, and on bandwidth requirements, we'll update this (as of 2013-09).

• How is financial data loaded into Checkbook?

  Right now data is loaded via an ETL (extract->transform->load) process, documented in Creating new Database and running ETL Job.

  The Checkbook repository provides some sample data to load, but note that the load process (the "TL" of "ETL") makes use of certain features of the Greenplum database. Once Checkbook is running, it does not require Greenplum; PostgreSQL will do fine. It's just the ETL process currently requires Greenplum. An enhancement ticket has been filed about removing this dependency: issue #19.

• Which side, export or import, is responsible for validity-checking and action determination in the ETL process?

  The original, expanded form of this question was:

  All the validity-checking and action determination is done in the transform/load steps, right? That is, the export side just exports everything -- it's a full dump, not an update -- and then Checkbook's transform/load process is responsible for determining valid/invalid, "I"nsert, "U"pdate, and "N"ame-change flags? (In other words, those flags are not in the data as delivered from FMS.)

  I think that's the case, from looking at the documentation and the test data in source/database/ETL/SOURCE_DATA/, but I wanted to make sure I understood completely.

  Checkbook developer Tirupati Reddy of REI Systems answered. Here is his answer, lightly edited for the FAQ:

  Loading to staging tables, performing validation, and processing data to transaction tables (whether to update/insert) is done in the Transform and Load steps. The source data is not always a full dump: we get the full dump every day only for the referenced elements like Departments, Expenditure Objects, etc, and Pending Contracts. It is incremental update for all other data like contracts, spending, payroll etc. Yes, the Insert/Update flags are determined in Process/Load step -- the export source does not provide those.