Update OpenMRS completely damaged my personal concept dictionary

Tags: #<Tag:0x00007fa3f4247490>

Hi Everyone.

In the last 4 days I shut down my Tomact server to do maintenance tasks in my OpenMRS implementation, I updated referenceaplication to 2.10, platform to 2.3 and some modules I had outdated, like addresshierarchy to 2.12.0 and htmlformentry to 3.10.0.

Checking that function by function everything was fine, I found that some forms gave me an error when opening them. I investigated my dictionary of concepts and, for some reason, 249 of my concepts have been updated by concepts of OpenMRS with Covid-19 data.

I understand that you want to help with this type of concepts to the current world situation, but with this you have also damaged 249 concepts that I have created during these 2 years, along with the HTML forms and reports attached to these concepts.

I don’t need help to restore them, fortunately I have a backup of the database and I am recovering the damaged, but, this kind of changes should be different. Add in another concept number that has not been used or at least a warning before updating.

Now I will be very careful with the updates, moreover, I think I will never update again.

i can identify all the concepts with changed by 1 and date_changed on march

I hope this serves as a warning to someone who has enough concepts in their dictionaries, like me.

1 Like

It looks like COVID-19 concepts were introduced to the Reference Metadata module.

@dkayiwa, am I correct in assuming the Reference Metadata module is meant for supplying default metadata for the Reference Application and/or running the Reference Application with demo data? Certainly, implementers upgrading their systems shouldn’t have Reference Metadata running and overwriting concept by internal concept ID.

/cc @mozzy, @herbert24

Yes you are correct.

The reference application has not yet played well with custom concepts. @willa got into some sort of a related problem here: Concept Dictionary Changes When Upgrading from 1.11.6 to Refapp 2.8.0

@ssmusoke uses some bit of hack to workaround it for UgandaEMR

1 Like

@burke @akanter This is a very bad decision IMO, the COVID concepts do not belong in Reference Application, rather they should be part of a custom module and distribution focused on the COVID response, otherwise it ties OpenMRS to COVID, yet there is more to OpenMRS base usage than COVID. This is a similar approach that I recommended for disease specific and vertical distributions on top of RefAPp

@elkinleon this will however happen whenever you upgrade Ref App due to the concepts that are dependent on the numeric IDs. I would suggest you increment the numeric IDs of all concept related tables by 5000000 (Five Million) before you start creating custom concepts.

I have a blog post coming this week that details this approach


thanks @ssmusoke, i will do that!

Thanks to @burke and @dkayiwa to clarify and help to understand what happened.

We definitely need a way to integrate OCL more into these releases so that folks have a more robust way to handle custom concepts and integration of CIEL/Ref App content…


I will bring this up in other venues as well (and I did already here on Slack), but IMO we need to work on deprecating the Reference Metadata module.

Just me thinking out aloud on OCL, would it not be more efficient to build a new toolchain from the ground up with the latest Python/Django as the current one seems to be dragging along

@elkinleon as promised An Opinionated Approach to OpenMRS Concept Management