[Feedback] Multiple Datasources

[kavilla]

OpenSearch Dashboards provides tools to help you visualize and explore your data. Multiple datasources project allows for the connection to an OpenSearch cluster endpoint

[kavilla]

Creating a data source connection says: A data source is an OpenSearch cluster endpoint (for now) to query against. I think we should drop the for now. It is kind of ambiguous and implies a guarantee that has not been verified as feasible.

[kavilla]

Data sources page reference legacy application

[abbyhu2000]

Maybe consider adding a readme doc to explain how to use the multiple datasource plugin and its functionalities?

[kavilla]

Unsure of the verification behind adding a data source connection. Still verifying. I am able to add a duplicate datasource will see the impact of this.

EDIT: By adding a duplicate data source it doesn't impact the viz's for the original data source. Experience is on par with deleting index pattern (all associated viz and dashboards remain but fail to work because they can't find index pattern)

[kavilla]

This will delete data source connections(s) and all Index Patterns using this credential will be invalid for access.

connections(s)

[kavilla]

I believe this not the standard from OpenSearch's for modal that can delete something. @KrooshalUX

[kavilla]

Sample data is able to add duplicate index patterns, should there be an indication on the index pattern page which source the index pattern is from?

Not sure if possible to do this without using the sample data stuff

Perhaps a bug improvement to have the sample data page check if MD is enabled and check if those patterns exist on each cluster to show if they should show the "add" button or "remove"

[kavilla]

Information how my plugin can extend this functionality, ie would be cool to add this feature to the console plugin.

[kavilla]

⚠️ for validation again, it does not do any validation on the version of the data source. currently we are coupled tightly with the version of OpenSearch. But it could get data from a data source that is really old or really new version with document's or APIs changes that we have yet to implement in the current version.

[kavilla]

⚠️ If MD is enabled, how did we want to handle if any outage on the other data sources? What is the experience? It seems like it will start since it will just check the default data source but visualizations associated to that index pattern from the data source should fail to load. Do we indicate that the data source is experiencing an outage or not found?

[kavilla]

Batching concurrent requests fails

[ananzh]

Maybe consider adding a readme doc to explain how to use the multiple datasource plugin and its functionalities?

Yes, I agree with @abbyhu2000. I have trouble to set it up as a developer. A developer guide / readme would be better with 1) setup in opensearch_dashboards.yml 2) instructions and example to create a data source 3) provide some sample endpoint_url

[ananzh]

I see the column sorting can do the sort on the lowercase named data source correctly. But if there is capital letter named data source or symbol started named data source, it is not working properly. Is there any restrictions on how to name the data source? is this some expected behavior?

[ananzh]

Same with description sorting. not sure whether this is expected. seems if cx doesn't pay attention to lowercase/uppercase, then cx might not be able to find the correct data source immediately if there are a lot. If this is expected, we need to notice cx in the user guide.

[ananzh]

I am reading this test:

    * Using the search bar, type the search criteria to find something that matches only 1 items in the table (search for title)
    * Validate that table displays only one row
    * Using the search bar, type the search criteria to find something that matches only 1 items in the table (search by description)
    * Validate that table displays only one row

but it seems we are allowing same data source (title, description, endpoint url and Authentication method) to be created

is this expected? from the test description, it seems not. I think this is nit, just curious whether this is what we expected.

[abbyhu2000]

When i try to update password, there is no indication on if i successfully updated the password or not.

1. If i provide a wrong password for Old password, there is no error shown and it still allows me to click the update button. The pop up is then closed and return to the previous page. It gives an illusion that the password is successfully updated with a wrong Old Password. (I think it's helpful to have some messages like "Please provide a correct old password", and update button should not be allowed)
2. If i provide a correct password for Old password, there is no message to indicate that I successfully updated the password.

[ananzh]

When I try to create an index pattern with system index. It is not showing properly.

1. search .kiban --> can see all the indices

2) search .kibana --> show only one. should this match all indices starting with .kibana?

then if you add _ it starts to show again

[ananzh]

should me remove alias?

[abbyhu2000]

After I created a new index pattern from an external data source, the View single document is broken.

View single document should work as below(this is the index pattern using the default data source):

[kavilla]

Explicit message indicating that this is an experimental feature for example like the visualization builder:

[KrooshalUX]

Moving a conversation from Slack into GitHub: There are still pending UX changes and copy edits to be made that had to be rescheduled before this merge was made. I will work with @kristenTian, @mpabba3003 and others to help address those feedback items.

[bjo004]

Maybe consider adding a readme doc to explain how to use the multiple datasource plugin and its functionalities?

Yes, I agree with @abbyhu2000. I have trouble to set it up as a developer. A developer guide / readme would be better with 1) setup in opensearch_dashboards.yml 2) instructions and example to create a data source 3) provide some sample endpoint_url

Hi @ananzh ,

Should I assume it's the opensearch URL that will look like this https://<OPENSEARCH_URL>.com:9200

Kind regards,

Bankole.

[zhongnansu]

⚠️ for validation again, it does not do any validation on the version of the data source. currently we are coupled tightly with the version of OpenSearch. But it could get data from a data source that is really old or really new version with document's or APIs changes that we have yet to implement in the current version.

Team decides it's not a release blocking issue, added to the roadmap of 2.5, #2435

[mpabba3003]

I see the column sorting can do the sort on the lowercase named data source correctly. But if there is capital letter named data source or symbol started named data source, it is not working properly. Is there any restrictions on how to name the data source? is this some expected behavior?

This is default sorting by EUI library InMemoryTable.

[zhongnansu]

Data sources page reference legacy application

I didn't get this one, do you have a screenshot? Is it about documentation links on the page? @kavilla

[zhongnansu]

Information how my plugin can extend this functionality, ie would be cool to add this feature to the console plugin.

#2437 #2436

[mpabba3003]

When i try to update password, there is no indication on if i successfully updated the password or not.

1. If i provide a wrong password for Old password, there is no error shown and it still allows me to click the update button. The pop up is then closed and return to the previous page. It gives an illusion that the password is successfully updated with a wrong Old Password. (I think it's helpful to have some messages like "Please provide a correct old password", and update button should not be allowed)
2. If i provide a correct password for Old password, there is no message to indicate that I successfully updated the password.

This is still under work. As of now nothing happens when user tries to update password

[zhongnansu]

Unsure of the verification behind adding a data source connection. Still verifying. I am able to add a duplicate datasource will see the impact of this.

EDIT: By adding a duplicate data source it doesn't impact the viz's for the original data source. Experience is on par with deleting index pattern (all associated viz and dashboards remain but fail to work because they can't find index pattern)

#2411

[zhongnansu]

Information how my plugin can extend this functionality, ie would be cool to add this feature to the console plugin.

It's on the roadmap. To support Legacy client in ds is paving the road for external plugin integration. I'll create an issue for tracking

[zhongnansu]

Batching concurrent requests fails

#2174 tracking here