Success

User tests: Successful: Unsuccessful:

avatar mbabker
mbabker
1 May 2015
  • Adds unique index to the #__issues table on issue number and project ID
  • Adds cascade operations to the #__issue_category_map table's foreign key to the #__issues table
avatar mbabker mbabker - open - 1 May 2015
avatar b2z
b2z - comment - 4 May 2015

So I suppose I cannot test it, right?

avatar mbabker
mbabker - comment - 4 May 2015

You could do it in a local database. If you delete a record from the #__issues table, the associated rows in the #__issue_category_map table should also delete with this patch (right now MySQL causes the delete to fail due to foreign key constraints).

For the unique index, apply it then try adding a record that has the same project ID and issue number as an existing row, it should fail. But you should be able to add several records with the same issue number and different project IDs.

avatar b2z b2z - reference | 292f9e0 - 5 May 15
avatar b2z b2z - merge - 5 May 2015
avatar b2z b2z - close - 5 May 2015
avatar b2z b2z - change - 5 May 2015
Status New Closed
Closed_Date 0000-00-00 00:00:00 2015-05-05 04:34:30
Closed_By b2z
avatar b2z b2z - close - 5 May 2015
avatar b2z
b2z - comment - 5 May 2015

All good :+1:

avatar b2z b2z - head_ref_deleted - 5 May 2015
avatar mbabker
mbabker - comment - 5 May 2015

11 issues failed the unique key index when this change was made on the live database. Wow...

avatar b2z
b2z - comment - 5 May 2015

Well that means we have some issues in our code. Do you have an info where does it happen?

avatar mbabker
mbabker - comment - 5 May 2015

Not a clue.

avatar b2z
b2z - comment - 5 May 2015

Sad... Is it on cron job right? May be we need to add more logging there.

avatar mbabker
mbabker - comment - 5 May 2015

The webhooks are received and processed in real-time. The cron jobs just make sure we stay in sync for changes that don't have hooks. I think the insert issue is probably coming from the hooks, I just have nothing to support that yet. I can probably rotate out all the logs on the server and enable the database logger, but that one writes every recorded query and I think causes us to hit our hard drive space limits in 5-7 days (in addition to the other loggers).

Add a Comment

Login with GitHub to post a comment