CoTwins

From Vrieze Wiki
Revision as of 14:34, 11 May 2020 by Scott (talk | contribs)
Jump to navigation Jump to search

Meeting minutes

Meeting minutes.

Software development tracking

In the initial NIAAA phase of CoTwins we worked with Anvil8, a software development company. At that time, we used agile software development principles and track our progress on https://youtracker.anvileight.com. This relationship ended with the end of the NIAAA support. Starting in

Software repo and bug reports

Software was originally maintained in bitbucket. Bug reports were recorded in BitBucket's issue tracker.

As of early 2020, the app engine software is maintained in GitLab. EF and iOS code are maintained at github.umn.edu.

EF Data Analysis

EF Data has been coded and analyzed through August 17th, 2016 (577 Subjects) by Daniel Gustavson. The EF task scripts, data, and most recent upload, can be found on the delta server (delta.colorado.edu\COTwins\EF Analyses). There are ReadMe files inside the folders for each of the four steps for data analysis (start there for instruction on how to add future subjects to the data), and an archive of Dan's last upload (delta.colorado.edu\COTwins).

Follow up note from Andrew Reineberg on 5/15/2017:

   I added a new folder to cotwins on delta (EF Analyses - 2017-05-13) containing the analyses using Dan's scripts as well as a python script for calculating descriptives and reliability (andrew_ef.{ipynb,html,pdf} inside the 4_Final EF Data directory). About 100 participants were added since the last analysis in August of '16 for a final n of 654 (cases where we have all three EF tasks). The reliability and interrelationship of EF tasks is consistent with the summaries Dan provided of previous time points. FYI, we noticed split half reliability of the task set switching task is low compared to other samples. One other oddity I found is the modification of participant 23521's keep track data file. I'll document the issue here in case you want to forward this to someone who can fill us in on the details:
   There appear to be several modifications of the file, so I am not sure which one is appropriate to use. All comparisons are to the raw data file (20160609ApaperdataCCW.xlsx).
   Mod 1 - 20160609ApaperdataCCW_cleanedlaterIS_MS.xlsx - incorrect participant number was changed to the correct participant number. Response to a practice trial was changed. These modifications should not affect keep track score. I used this file in my analysis.
   Mod 2 - 20160609ApaperdataCCW_cleanedlater_IS_MS.xlsx - incorrect participant number was changed to the correct participant number. Response to the first real trial was changed. The latter modification will affect keep track score.
   At this point, the EF variables can be put into whatever higher level analyses you had in mind. I suppose that is something you'll be discussing with Naomi. Do let me know if there is anything I can do to facilitate those analyses or access to the work Dan and I did.
  -Andrew"

Changing emails in production

Email from Paul Resnick in December 2017 describing how to update an email address in the production database.

Here's what I did, in case someone else ever has to do this again. But be very careful; you could really do damage if, for example, you left off the where clause from the SQL update queries.

1) First, I went to the google cloud console, for the project Co-twins Users, https://console.cloud.google.com/sql/instances/user-db/overview?project=co-twins-users

2) Go to the SQL (under storage) on the main menu

3) Click on the only instance, user-db

4) Click on "Connect Using Cloud Shell". Now you'll get a unix command prompt, with a command line ready to run. But you will want --user=admin rather than --user=root. So, something like this:

  presnick@co-twins-users:~$ gcloud beta sql connect user-db --user=admin

For password, you'll need to get it from the password file. It's the thing after mysql://admin: and before the @

5) Now you'll be at an SQL prompt. I entered the following:

use user_schema;

select * from users where email_id = 'original_email@original_email.com';
/*(should get back 1 row for a correct email address)*/

update users set email_id = 'new_email@email.com' where email_id = 'original_email@original_email.com';
/*(changes all rows for that email address, in our case just 1)*/

Updating the openshift server that serves the EF tasks

In December 2017 Alejandro dlV changed the hosting server for the EF tasks from an old openshift platform to a new one in response to some major changes in OpenShift products. On Dec 26th, Paul R. made the following change to the Co-Twins server to point to the new openshift instance.

<---- Begin Paul's Email ---->

I have completed this now.

Here's what I did:

  • logged into CloudSQL instance via google cloud console, as per password change I made previously
  • ran the following SQL queries to change all not-yet-completed surveyentries to point to the new URL:
update surveyentries set url = 'http://assesment-cotwins.b9ad.pro-us-east-1.openshiftapps.com/exp/task/keep_track' where url = 'http://assesment-cotwins.rhcloud.com/exp/task/keep_track' and completed = 0;
update surveyentries set url = 'http://assesment-cotwins.b9ad.pro-us-east-1.openshiftapps.com/exp/task/category_switch' where url = 'http://assesment-cotwins.rhcloud.com/exp/task/category_switch' and completed = 0;


  • updated surveyconfig.xml so that new entries would be created using the new url from here on out
  • tested on my test user account
    • the existing surveyentry points to the new server and works
    • deleted it and recreated it; the new url points to the new server and also works

<---- End Paul's Email ---->

Feb 2018 userStatus and survey problems

Around Feb 2018 some participants began having trouble taking surveys and, at the same time, userstatus became too slow to be able to track surveys. This problem was fixed after 2-3 weeks, or so, with the following changes:

1) There were a *lot* of broken survey tokens caused by a problem in the limesurvey tokens DB. I fixed these using the fix feature that Vitaliy implemented the last time we had a problem like this. You should monitor for any similar problems for the next few days, to see if it recurs. To do that, visit the following URL: https://co-twins.appspot.com/admin/broken-survey-tokens?limit=10000&start=20000

2) The user-status admin page had become too slow because of the query to find the last web page visited for every user. We removed that information from the interface and now the page loads pretty quickly.

April 2018 SSL certificate authority chain problem on android devices

When we put the new SSL cert on the server, he was supposed to update the "certificate authority chain" as well. Most browsers know about the certificate authority I'm using, and didn't need to check that chain, but apparently not the Android browser.

April 29 2020 iOS update

A refactored version of the iOS app, with essentially the same functionality as the previous app but rewritten in swift programming language, was release to the Apple app store. Scott (talk) 12:15, 30 April 2020 (CDT)