You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@nifi.apache.org by "John Wise (Jira)" <ji...@apache.org> on 2022/03/31 19:45:00 UTC
[jira] [Updated] (NIFI-9708) Add "lazy loading" to NiFi Registry
[ https://issues.apache.org/jira/browse/NIFI-9708?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
John Wise updated NIFI-9708:
----------------------------
Description:
We have a couple dozen buckets, containing 100+ versioned flows, all stored in git. When the Registry's user interface starts, it currently loads all of that information before the UI completes rendering. The initial UI (banner, user info & sort/filter bar) starts to render at ~12 seconds, but the whole page doesn't completely render for ~30 seconds. Even when that's done, open & closing a flow div takes quite a while because it has to iterate through all of the flow information that's now stored in JavaScript memory.
Ideally, the UI should initially just display the list of buckets, preferably in tree format. When the user clicks on a tree node, the flow information for that bucket would be loaded via JS. Clicking a particular flow entry would then load the remaining information for that flow. That would improve startup & flow navigation time immensely.
Filtering would obviously require additional JS loading to complete.
Also, there's a ton of wasted screen resolution in general in the UI, which could be used for things like the flow thumbnails others have asked for, and/or multiple flow columns - e.g.
|[img] flow |[img] flow |[img] flow |
|[img] flow |[img] flow |[img] flow |
|[img] flow |[img] flow |[img] flow |
instead of just
|flow |
|flow |
|flow |
was:
We have a couple dozen buckets, containing 100+ versioned flows, all stored in git. When the Registry's user interface starts, it currently loads all of that information before the UI completes rendering. The initial UI (banner, user info & sort/filter bar) starts to render at ~12 seconds, but the whole page doesn't completely render for ~30 seconds. Even when that's done, open & closing a flow div takes quite a while because it has to iterate through all of the flow information that's now stored in JavaScript memory.
Ideally, the UI should initially just display the list of buckets, preferably in tree format. When the user clicks on a tree node, the flow information for that bucket would be loaded via JS. Clicking a particular flow entry would then load the remaining information for that flow. That would improve startup & flow navigation time immensely.
Filtering would obviously require additional JS loading to complete.
Also, there's a ton of wasted screen resolution in general in the UI, which could be used for things like the flow thumbnails others have asked for.
> Add "lazy loading" to NiFi Registry
> -----------------------------------
>
> Key: NIFI-9708
> URL: https://issues.apache.org/jira/browse/NIFI-9708
> Project: Apache NiFi
> Issue Type: Improvement
> Components: NiFi Registry
> Reporter: John Wise
> Priority: Minor
>
> We have a couple dozen buckets, containing 100+ versioned flows, all stored in git. When the Registry's user interface starts, it currently loads all of that information before the UI completes rendering. The initial UI (banner, user info & sort/filter bar) starts to render at ~12 seconds, but the whole page doesn't completely render for ~30 seconds. Even when that's done, open & closing a flow div takes quite a while because it has to iterate through all of the flow information that's now stored in JavaScript memory.
> Ideally, the UI should initially just display the list of buckets, preferably in tree format. When the user clicks on a tree node, the flow information for that bucket would be loaded via JS. Clicking a particular flow entry would then load the remaining information for that flow. That would improve startup & flow navigation time immensely.
> Filtering would obviously require additional JS loading to complete.
> Also, there's a ton of wasted screen resolution in general in the UI, which could be used for things like the flow thumbnails others have asked for, and/or multiple flow columns - e.g.
> |[img] flow |[img] flow |[img] flow |
> |[img] flow |[img] flow |[img] flow |
> |[img] flow |[img] flow |[img] flow |
> instead of just
> |flow |
> |flow |
> |flow |
--
This message was sent by Atlassian Jira
(v8.20.1#820001)