Skip to content

Historic: Testing the 2012 Database Schema Conversion

Robert Sparks edited this page May 1, 2023 · 1 revision

This page is out-of-date; the database schema conversion was completed in January 2012


Instructions for testing of the DB Schema Conversion Release

The following instructions are applicable for Phase 3 of the testing of the database schema conversion datatracker release, scheduled for February 2012.

Background

We are now getting close to a point with the IETF datatracker which we have been working on for something like 3 years:

The deployment of a new database schema which has been designed for the tasks we are using it for today (with some thoughts for tomorrow) rather than the old schema which was good when created more than 10 years ago, but which we've outgrown a long time ago.

Testing Phases

The testing is planned to take place in 3 phases over the next 4 weeks (from January 16th to February 17th). In each phase, I would like you spend some time doing the things you usually do with the datatracker.

  • Phase 1 is a sanity check case. During this, the code is running in an environment where it can be very easily updated with bugfixes, but is single-threaded and can't take much load. Here the code is running against a stable database snapshot which will not be up-to-date.

  • Phase 2 is for more thorough testing. During this, the code will be running on one of the tools server, under Apache, deployed in the same manner it will have in production use. This makes code updates more cumbersome, but it will be able to take more load, and it will run against a database which is updated hourly. This phase is when the bulk of your testing would be needed. Hopefully you can try out most of your different tracker use cases during this time.

  • Phase 3 is for wider community testing. During this, the code will be running on the production server, and we will try to make changes in the production database show up in the test tracker with very little delay. During this phase, it should be possible to use the test tracker instead of the production tracker for all read-only tasks. Tasks where changes are entered into the test database will however also have to be done in the production tracker; no changes will flow from the test tracker to the production tracker.

Testing Timeline

The following timeline is tentative; slippages may occur::

Jan 16 Phase 1 testing starts Jan 16-22 Fixing issues as testing goes on Jan 16-22 Sign up more testers Jan 26 Phase 2 deployment on tools server, with hourly DB sync and iterative conversion running Jan 30 Phase 2 testing starts Feb 3 Phase 2 testing ends ? Feb 6-11 Phase 3 deployment on main ietf server, with hourly? iterative conversion from live database, and tweak of the production tracker as needed Feb 14 Phase 3 testing starts Feb 17 Phase 3 testing ends ? Feb 15 Go/No Go decision Feb 18 Production cut-over ?

Phase 3 Server Setup

This deployment permits you to log in with your regular datatracker login, and do everything you would normally do in the tracker. Any changes you commit will however not be propagated to the production tracker, and may be over- written by data from the production tracker if changes to the equivalent page is made there.

The Phase 3 deployment is currently getting new document data from the production tracker every 5 minutes, so in this phase, you should expect document data to be up-to-date and correct unless the change was very recent.

Data which isn't document-related is converted from the production system once per 24 hours, at approximately 01:05 - 02:40 PST.

The Phase 3 deployment is running on the same server and with the same Apache configuration as the production server, so you should expect similar response times as with datatracker.ietf.org.

Please bang on this test instance by doing the kind of things you normally do with the datatracker, and let us know on the mailing list what you find. Your efforts will be much appreciated.

Gettin Email from the Test Server

Since the test server is running in test mode, all email to be sent out is intercepted before it actually goes out, preventing delivery to the address that would normally receive it.

If you're testing functionality which should result in emails going out, you'd still like to see the mails which would have been sent, to verify that things work as intended, however. In order to make this possible, we've set up an exception mechanism for email messages:

To receive a copy of each message before it's discarded, you can set up an email address to which delivery will be done anyway. The address is stored in the testmailcc cookie in your browser, so make sure cookies are enabled.

To set up an address for delivery from the test server, please go here: http://trackerbeta.ietf.org/accounts/testemail/

Starting Point

The Phase 3 test deployment is available here: http://trackerbeta.ietf.org/

Issue Reporting

Please report issues to the mailing list iola-conversion-tool@ietf.org.

Clone this wiki locally