Health Endpoint Design Document

[sezal98]

Why make this change?

Design Document for Health Endpoint
Resolves: 2504

What is this change?

Adding a design doc to help contributors to get an understanding of how the end to end health endpoint is coded and tested

• Added coding specifics, functions and conditions
• Added Test scenarios

[abhishekkumams]

How would the health endpoint will function with hot reload? In case of bad config it might use last good known config which might give wrong status.

Would be good to add limitations.

[sezal98]

How would the health endpoint will function with hot reload? In case of bad config it might use last good known config which might give wrong status.

Would be good to add limitations.

I dont see an issue in this, If DAB Engine is working then health point should be consistent with that. If DAB engine is using the last known config, then health endpoint would also show the health of the last known config.
Showing health of the current config but running the engine with the last known config would create inconsistencies which would be incorrect for the user.
We should put this in documentation that we are using the last known config in case of hot reload and showing health of that old config not the bad config.

Another important point to note is.. Health is independent of Validity of config. Health endpoint is only shown when the config is valid (no errors) and DAB engine is running. So the case of bad config would not occur.

@JerryNixon , Any thoughts on this?

[abhishekkumams]

How would the health endpoint will function with hot reload? In case of bad config it might use last good known config which might give wrong status.
Would be good to add limitations.

I dont see an issue in this, If DAB Engine is working then health point should be consistent with that. If DAB engine is using the last known config, then health endpoint would also show the health of the last known config. Showing health of the current config but running the engine with the last known config would create inconsistencies which would be incorrect for the user. We should put this in documentation that we are using the last known config in case of hot reload and showing health of that old config not the bad config.

Another important point to note is.. Health is independent of Validity of config. Health endpoint is only shown when the config is valid (no errors) and DAB engine is running. So the case of bad config would not occur.

@JerryNixon , Any thoughts on this?

Thanks for clarifying. can you add the same to your documentation as well.

[abhishekkumams]

How will we be checking if an endpoint is healthy or not? Are we gonna make a db call internally? Or hit the endpoint itself to check the status? Or something else?

[sezal98]

How will we be checking if an endpoint is healthy or not? Are we gonna make a db call internally? Or hit the endpoint itself to check the status? Or something else?

Yes, as mentioned in the document, we would be making the call internally and hitting the specified query. If this query is sucessfull and within the threshold specified, the entity would be considered healthy. Else, we would say, its unhealthy with the exception message

[Aniruddh25]

How will we be checking if an endpoint is healthy or not? Are we gonna make a db call internally? Or hit the endpoint itself to check the status? Or something else?

Yes, as mentioned in the document, we would be making the call internally and hitting the specified query. If this query is sucessfull and within the threshold specified, the entity would be considered healthy. Else, we would say, its unhealthy with the exception message

What if the specified query itself has Syntax errors? if the datasource is healthy, we might report a false -ve. how do we plan to validate the query text?

[Aniruddh25]

Fix the PR description to remove the text from the PR template

[Aniruddh25]

Waiting for addressing the comments

[JerryNixon]

@Aniruddh25

[sezal98]

/azp run

[abhishekkumams]

LGTM, thanks for updating.

[sezal98]

/azp run

[sezal98]

/azp run

[sezal98]

/azp run

[Aniruddh25]

This has come a long way! Thank you for your patience in addressing all the comments.

Left a few comments to answer but other than that looks good to merge.

[sezal98]

/azp run

[sezal98]

/azp run

Merging for now. Will address any more comments in further PRs

[sezal98]

/azp run