You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ozone.apache.org by "Abhishek Pal (Jira)" <ji...@apache.org> on 2023/08/25 11:34:00 UTC

[jira] [Closed] (HDDS-9124) [Recon] Network requests not canceled when context changes

     [ https://issues.apache.org/jira/browse/HDDS-9124?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Abhishek Pal closed HDDS-9124.
------------------------------

> [Recon] Network requests not canceled when context changes
> ----------------------------------------------------------
>
>                 Key: HDDS-9124
>                 URL: https://issues.apache.org/jira/browse/HDDS-9124
>             Project: Apache Ozone
>          Issue Type: Improvement
>          Components: Ozone Recon
>            Reporter: Abhishek Pal
>            Assignee: Abhishek Pal
>            Priority: Major
>              Labels: pull-request-available
>         Attachments: HDDS-9124.mp4
>
>
> With the current code for Recon UI, we are not handling scenarios where network requests should be cancelled when user is shifting pages/context.
> This causes unnecessary requests to be made.
> For example if the user opens the Recon UI and selects a view, then the user changes that view immediately, the request for the previous view will still go through, along with the new request for the changed view.
> This is more noticeable in a slow network, where requests might be stuck in a pending state, and during that time if we switch the view the previous request will still be sent, along with the new request.
> The scenario is attached in the video recording, where a slow 3G network is being emulated.
> If we are clicking on DiskUsage view it sends a request to:
> {panel} .../du?path=/&files=true {panel}
> Now if we click on the Containers view the second request is being sent for:
> {panel}.../unhealthy {panel}
> At this point we will notice both the requests to be in a pending state. And later both the requests are resolved.
> Ideally we should be cancelling the DiskUsage request the moment we shift to the container view.
> ----
> Note: This might not be affecting in an environment where users might be going through the data in Recon and they are expected to not cancel/change views immediately. But there might still be scenarios where the users click on the wrong view and later click on the correct option.
> For a small number of requests it is acceptable, but as the codebase grows and there are several features, mixed along with increased latency in network, this might become a problem and use up resources to serve requests that are no longer required.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@ozone.apache.org
For additional commands, e-mail: issues-help@ozone.apache.org