h4. Community Options
* https://github.com/CrunchyData/postgres-operator
* https://github.com/zalando/postgres-operator
h4. What option from Community shows better and why?
* https://github.com/CrunchyData/postgres-operator
Why:
* Better doc
* This one is in the [operatorhub.io|https://operatorhub.io/?category=Database] and the other not
h4. What image and version CrunchyData?
* Shows that it will use your on images
* Maybe it can be possible b changed. See [CCPImageTag|https://crunchydata.github.io/postgres-operator/latest/configuration/pgo-yaml-configuration/]. But we need to check if it actually works and how to setup the args used since each image has your own name of parameters to be informed. (E.g the key of the environment variable to define the DB user is not the same between the redhat image and the PostgreSQL official image.
h4. What image and version we are using currently?
* The imaged used in this project is from Red Hat. See more in https://docs.okd.io/latest/using_images/db_images/postgresql.html
* image: "centos/PostgreSQL-96-centos7"
* The MSS project was developed and tested so far with PostgreSQL 9.6
h4. Pros and Cons of change the current implementation for it
||PROS (+)||CONS (-) ||
|Has a bkp feature implemented|It has many features and we need to understand how to set up , configure and use|
|Show as a mature project that could be used|Show that we will need to use their images and not the RedHat ones, however, shows that their image is based on centos |
|Big documentation|We may face some issues and limitations as for example, we will be not able to do some specific setup and or fix some issue. |
| | We need to spend an effort to try to make it work as a dependency and handled by the operator for 3.11 mini shift and in OCP 4|
| | May it will increase the complexity unnecessarily, currently |
Conclusion: IMHO we should keep the database as a type for now and do not care over at least currently. Following the reasons
* The current implementation attend all our needs and is quite simple and easy to keep maintained, changed, replaced and or disabled.
* Also, the current implementation allows we easily without any code change make the MSS use another database and/or disabled this type, change the image and/or its versions as the parameters and values used
* IMHO add now the CrunchyData shows spend effort unnecessarily and could make it more complex
|
|
