Cannot access dataflint UI through Knox gateway URL using the

[ef1236]

The bug is the same as described in [https://github.com//issues/13](This bug report)

When I try to access data flint through a knox gateway, it tries to access without the gateway

GET https:///api/v1/applications/<application_id/1/environment 404 (Not Found)

it should be call to: https:///gateway//sparkhistory/history/<application_id>/1/environment/

It looks like version 0.2.5 should have fixed it but I still encounter the same problem

[menishmueli]

@DanielAronovich please take a look, isn't it the same path you simulated with a proxy 1:1?

[DanielAronovich]

@ef1236, you are corret and it should have been solved.

I am on it.

Does the history server works well and the dataflint tab works well, but only when pressing "TO HISTORY SERVER" it breaks?

Does the "To spark UI" button on top of it works well?

Thanks!

[ef1236]

@DanielAronovich Hey thanks for the quick reply

We are using Knox With Cloudera.
The history server dataflint tab only works when I don't go through the Knox gateway.

When I access the sparkhistory server directly using the hostname and port:
Hostname:18489/history/application_/dataflint it queries for the environment endpoint correctly:
Hostname:18489/api/v1/applications/application_/environment

But when I access through the sparkhistory server, the URL is correct but the page tries to query for the environment endpoint with the wrong URL.
When I go through the Knox gateway - exapmle.example/gateway/cdp-proxy/spark3history/history/application_/dataflint,
It queries
exapmle.example/api/v1/applications/application/environment
instead of
exapmle.example/gateway/cdp-proxy/spark3history/api/v1/applications/application/environment
which would have worked (I queried it myself).

[menishmueli]

Wrote a fix and released version 0.2.6. I believe it will work this time
You can look at the fix here: 4786a1b

@ef1236 let me know if version 0.2.6 fixed the problem so we can close this issue

[ef1236]

Hey, 0.2.6 did not fix the problem. But I think I found the problem in the code
In spark-ui/src/utils/UrlUtils.ts in Line 38 the pathToRemove variable should remove the ".*" at the start because then the regex captures the whole start of the URL and removes also the gateway

// old
const pathToRemove = /.*\/history\/[^/]+\/dataflint\/?$/;
// new
const pathToRemove = /\/history\/[^/]+\/dataflint\/?$/;

Then it only captures the part from the /history and not the whole pathname

I Created a pull request with the change