Multi-tier deployment in a single environment - let apps automatically know which datasources to use!
Major development is often split into several environment or stages, in order to separate development (where changes are made) and production (where end users are). It is used as a phased approach of software testing and deployment, where each tier closer to the end-user represents more critical infrastructure.
However, some software requirements use a single environment for two different stages of a development. This can be due to license matters, hardware costs etc.
Say you have both DEV and TEST datasources accessible in the same Qlik Sense environment, how can you easily handle two streams for each purpose?
In the script you can easily define e.g.:
SET vG_Environment = 'dev';
which then results in:
This would allow the developer to manually test and switch different data-sources depending on where in the development stage he/she is testing. But how do you do this automatically during the different deployment stages? And if there is a third, separate environment (which perhaps the developer isn't allowed to access), how can you let the administrator save time by not manually editing every single load script?
The solution is to let the script connect to the internal REST connector and read what type of Stream the app is published to!
The stream-type is defined through custom properties.
This will enable apps automatically sense which environment-stage they currently reside in. This eases app transfers considerably.
It will also save tons of time.
Put a copy of the developer code (e.g. Developer code to copy.txt) in a shared network folder which could contain both the script and the QVD files to read. The folder is a read-only folder and accessible for all users within the Qlik Sense Data Load editor.
It works by reading which streams exists and if they have a DEV or TEST StreamType custom property assigned. It then maps the streams to where the app is published and assumes that there is from where it should reload. We use a REST connector to read from the repository itself towards qrs_apphublist and qrs_stream. Other APIs can be used as well. The repository is a PostgreSQL database.
Here is how to accomplish this...
For the administrator:
In the below scenario we have two machines:
Machine 1 - acting as a DEV and TEST machine. This machine is assumed below unless otherwise stated, as on #6.
Machine 2 - PROD
1. Adding REST connectors
Add the following internal connectors and make sure they are accessible and added, for administrators:
You can imitate the other built-in REST connectors properties or just add the following properties:
Auto-detect response type: Yes
Key Generation strategy: Sequence ID
Use Windows Authentication: Yes
Skip Server Certificate Validation: Yes
Use Certificate: No
Pagination type: None
Query parameters: xrfkey: 0000000000000000
Query headers: X-Qlik-XrfKey: 0000000000000000; User-Agent: Windows
Have a look at the help site on how the APIs work, in case of questions. Yes, qrs_app can be used, but as we will extensively use these we want to be able to configure, troubleshoot and use our very own connector.
Remember to deny access for anyone else than administrators.
2. Add a new Custom Property
Add a new custom property called "StreamType".
Add two values: dev, test
3. Add the custom property values to the relevant streams
Apply these two values to the relevant streams, e.g.
StreamType - "dev"
StreamType - "test"
4. Create a new app, with the sole purpose of extracting Stream ID and corresponding custom properties.
Create a new app which contains (code at "Streams to QVD.txt"), which extracts which Stream IDs that has the corresponding custom properties and saves this to a file. This will use our REST connector "qrs_stream".
As streams are not as often added it should be enough to run once a day, so remember to schedule it.
The reload task should output a QVD file called "stream_id_to_env.qvd".
5. Make the developers code easily accessible
Now, put the code attached (whichenvironment.txt) in a text-based file which is accessible (read-only) for the developers, e.g. the "Services" library.
This folder needs to be a shared network drive, so that the developer can easily copy the file content.
Remember to add the folder as a Library as well. The library should be accessible to everybody in order for the developer to read the "stream_id_to_env.qvd" from his script.
In the DEV/TEST environment our apps will read this text file and set the corresponding variable depending on which stream the app is published to, by connecting to the internal QRS database and read which Stream it is published to. If the app contains "_DEVELOPMENT" in the title, the rest is ignored and "DEV" is assumed, which allows for the developer to use the include-file despite not being able to access the qrs_stream and qrs_apphublist data-sources. It thus just skips the rest of the script.
The app then knows if it is to access "dev" or "test" data sources.
Developers can also manually define which connection to use, during testing.
6. Repeat the above steps in any other environment
In another environment (e.g. PROD) where there is only a single tier, the whichenvironment.txt can be very simple and include a simple variable "SET" for the app to read, in order to know which datasource to use.
Remember to share the network folder in PROD as well.
For the developer:
Include the following in the script:
//Feature to sense which environment we are working in