HDDS-11150. Recon Overview page crashes due to failed API Calls by spacemonkd · Pull Request #6944 · apache/ozone

spacemonkd · 2024-07-15T06:57:20Z

What changes were proposed in this pull request?

HDDS-11150. Shift Axios.all to Promise.allSettled and add fallback for error in failed requests

Please describe your PR in detail:

Currently Axios.all() uses Promise.all() in the background, but this causes all further promises to reject if one request is rejected
So we have migrated to Promise.allSettled() which allows to wait for all requests to complete instead of immediately rejecting
Along with this new fallbacks have been added to allow data to be shown even if there is an error while fetching the data
Scope of fallback scenario
- When the API is unreachable, say because of an API timeout or the backend is not able to respond
- When the network might go down and the call to the backend fails
- If the API returns any error (like 500, or 404 etc.) where the data is not present

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-11150

How was this patch tested?

Patch was tested manually

Screenshot	Description
	Fallback when clusterState request fails
	Fallback when open keys summary fails
	Fallback when deletePending keys summary fails
	Fallback when File Size Distribution API fails
	Fallback when Container Size Distribution API fails

… fetching data

spacemonkd · 2024-07-15T06:59:23Z

@ArafatKhan2198 @devmadhuu @smitajoshi12 Could you please take a look?
Also please do suggest if you have any alternatives to the ?? that I am displaying when the clusterState endpoint fails

devmadhuu · 2024-07-15T09:37:50Z

@devabhishekpal Can you pls define clear scope and clear boundaries this PR will cover ?

dombizita · 2024-07-15T14:07:29Z

@ArafatKhan2198 @devmadhuu @smitajoshi12 Could you please take a look? Also please do suggest if you have any alternatives to the ?? that I am displaying when the clusterState endpoint fails

I'd suggest to go with N/A, as that describes maybe better that we were not able to get the necessary data to display. Let me know what you think of this!

spacemonkd · 2024-07-15T16:31:43Z

Message displayed when we are hovering over the data in case of fallback.

smitajoshi12 · 2024-07-16T12:04:57Z

@devabhishekpal
Can we do same fall back for Open keys Summary and Pending Deleted Keys Summary for NA instead of 0.Not able to differentiate between 0 because of API Response 0 keys or it is because of fallback.

smitajoshi12 · 2024-07-16T12:17:47Z

@devabhishekpal

Can we handle null check before reduce

smitajoshi12 · 2024-07-16T12:54:39Z

@devabhishekpal

Same Issue with Count Response. please check API failing test case with wrong api name and async await.

…hts .reduce() error

spacemonkd · 2024-07-17T05:37:25Z

Hi @smitajoshi12 , thanks for your review.
Addressed the issues you pointed out in the latest commit

smitajoshi12 · 2024-07-17T07:32:16Z

@devabhishekpal

This condition should be in or
If we fail one api other api is not working for Insights
resp.status === 'rejected' || resp.reason.toString().includes('CanceledError')

spacemonkd · 2024-07-17T08:41:00Z

Hi @smitajoshi12, regarding:

@devabhishekpal
This condition should be in or If we fail one api other api is not working for Insights resp.status === 'rejected' || resp.reason.toString().includes('CanceledError')

Yes, but the reject will only happen when we completely shift to a different navigation/page.
At that point having the response for one of the endpoints is not required/not useful as the user will have already shifted to a different page

Shifting between FileSize view tab and the ContainerSize tab will not trigger the reject - so the data will get loaded properly.

spacemonkd · 2024-07-17T08:52:03Z

Extending on the above, if we use || it means the error would be thrown in either if it gets rejected, or if it is cancelled.
But this is not correct, as there might be scenarios where the API might reject the Promise, but the error is not cancellation.

In that scenario we do not want to throw the error, but fallback to the default values.
We want to throw the error only when the request is rejected as the call was cancelled.

smitajoshi12 · 2024-07-17T09:38:18Z

@ArafatKhan2198 @devmadhuu

Can you review as this pr needs to get merged in OZONE-719.3.x as we have already completed all recent changes cherry picks for ECharts and other upstream Jiras.

smitajoshi12 · 2024-07-17T09:47:40Z

Extending on the above, if we use || it means the error would be thrown in either if it gets rejected, or if it is cancelled. But this is not correct, as there might be scenarios where the API might reject the Promise, but the error is not cancellation.
In that scenario we do not want to throw the error, but fallback to the default values. We want to throw the error only when the request is rejected as the call was cancelled.

If one api fails because Internal server error or any other issue response is blank but UI looks this way It should catch error or throw error

spacemonkd · 2024-07-17T12:32:40Z

Thanks @smitajoshi12 for pointing this out.
Addressed this in the latest PR.

In case the request is cancelled, we go to catch.
Else we initialize with the default values and display a popup to notify the user something is going wrong.

Can you take another look?
Thanks

smitajoshi12 · 2024-07-18T07:01:40Z

It looks good to me

ArafatKhan2198

Thanks for the changes @devabhishekpal after discussing in our offline sync the changes LGTM!

smitajoshi12

All changes looks good to me

ArafatKhan2198 · 2024-07-19T05:42:32Z

Thanks for the patch @devabhishekpal and thanks for the review @smitajoshi12

…-delete * HDDS-10239-container-reconciliation: (184 commits) HDDS-10373. Implement framework for capturing Merkle Tree Metrics. (apache#6864) HDDS-11188. Initial setup for new UI layout and enable users to switch to new UI (apache#6953) HDDS-11120. Rich rebalancing status info (apache#6911) HDDS-11187. Fix Event Handling in Recon OMDBUpdatesHandler to Prevent ClassCastException. (apache#6950) HDDS-11213. Bump commons-daemon to 1.4.0 (apache#6971) HDDS-11212. Bump commons-net to 3.11.1 (apache#6973) HDDS-11211. Bump assertj-core to 3.26.3 (apache#6972) HDDS-11210. Bump log4j2 to 2.23.1 (apache#6970) HDDS-11150. Recon Overview page crashes due to failed API Calls (apache#6944) HDDS-11183. Keys from DeletedTable and DeletedDirTable of AOS should be deleted on batch operation while creating a snapshot (apache#6946) HDDS-11198. Fix Typescript configs for Recon (apache#6961) HDDS-11180. Simplify HttpServer2#inferMimeType return statement (apache#6963) HDDS-11194. OM missing audit log for upgrade (apache#6958) HDDS-10389. Implement a search feature for users to locate open keys within the Open Keys Insights section. (apache#6231) HDDS-10561. Dashboard for delete key metrics (apache#6948) HDDS-11192. Increase SPNEGO URL test coverage (apache#6956) HDDS-11179. DBConfigFromFile#readFromFile result of toIOException not thrown (apache#6957) HDDS-11186. First container log missing from bundle (apache#6952) HDDS-10844. Clarify snapshot create error message. (apache#6955) HDDS-11166. Switch to Rocky Linux-based ozone-runner (apache#6942) ...

spacemonkd added 2 commits July 15, 2024 12:15

Migrate AxiosAll to PromiseAllSettled and added fallback for error in…
f213df7
… fetching data

Added undefined check for summary data
dbdd378

spacemonkd changed the title ~~HDDS-11150. Shift Axios.all to Promise.allSettled and add fallback for error in failed requests~~HDDS-11150. Recon Overview page crashes due to failed API CallsJul 15, 2024

adoroszlai added recon UI labels Jul 15, 2024

Added error boundaries and changed message to N/A in case of API failure
84e5861

removed console.log lines
719aa34

Made changes to give N/A for summary endpoints and also handled insig…
e89c747
…hts .reduce() error

spacemonkd added 2 commits July 17, 2024 17:21

added an Error wrapper for all errors in Promise
ea2e800

Added error message for other axios errors
ae3dad1

spacemonkd added 2 commits July 17, 2024 22:01

Fixed indentation
6994a04

Added No data overlay for error in network response
5f59b52

ArafatKhan2198 approved these changes Jul 18, 2024
View reviewed changes

smitajoshi12 approved these changes Jul 19, 2024
View reviewed changes

ArafatKhan2198 merged commit 0c924bc into apache:masterJul 19, 2024

spacemonkd deleted the HDDS-11150 branch July 19, 2024 06:44

Uh oh!

Conversation

spacemonkd commented Jul 15, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What changes were proposed in this pull request?

What is the link to the Apache JIRA

How was this patch tested?

Uh oh!

spacemonkd commented Jul 15, 2024

Uh oh!

devmadhuu commented Jul 15, 2024

Uh oh!

dombizita commented Jul 15, 2024

Uh oh!

spacemonkd commented Jul 15, 2024

Uh oh!

smitajoshi12 commented Jul 16, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

smitajoshi12 commented Jul 16, 2024

Uh oh!

smitajoshi12 commented Jul 16, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

spacemonkd commented Jul 17, 2024

Uh oh!

smitajoshi12 commented Jul 17, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

spacemonkd commented Jul 17, 2024• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

spacemonkd commented Jul 17, 2024

Uh oh!

smitajoshi12 commented Jul 17, 2024

Uh oh!

smitajoshi12 commented Jul 17, 2024

Uh oh!

spacemonkd commented Jul 17, 2024

Uh oh!

smitajoshi12 commented Jul 18, 2024

Uh oh!

ArafatKhan2198 left a comment

Choose a reason for hiding this comment

Uh oh!

smitajoshi12 left a comment

Choose a reason for hiding this comment

Uh oh!

ArafatKhan2198 commented Jul 19, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

spacemonkd commented Jul 15, 2024•
edited
Loading

smitajoshi12 commented Jul 16, 2024•
edited
Loading

smitajoshi12 commented Jul 16, 2024•
edited
Loading

smitajoshi12 commented Jul 17, 2024•
edited
Loading

spacemonkd commented Jul 17, 2024•
edited
Loading