[SPARK-21254] [WebUI] History UI Performance fixes #18777
Closed
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…ssing Currently all the DOM manipulations are handled in a loop after Mustache template is parsed. This causes severe performance issues especially within loops iteration over thousands of (attempt/application) records and causing all kinds of unnecessary browser work: reflow, repaint, etc. This could be easily fixed by preparing a DOM node beforehand and doing all manipulations within the loops on detached node, reattaching it to the document only after the work is done. The most costly operation in this case was setting innerHTML for `duration` column within a loop, which is extremely unperformant: https://jsperf.com/jquery-append-vs-html-list-performance/24 While duration parsing could be done before mustache-template processing without any additional DOM alteratoins.
Check whether to display pagination or not on large data sets (10-20k rows) was taking up to 50ms because it was iterating over all rows. This could be easily done by testing length of array before passing it to mustache.
Logic related to `hasMultipleAttempts` flag: - Hiding attmptId column (if `hasMultipleAttempts = false`) - Seting white background color for first 2 columns (if `hasMultipleAttempts = true`) was updating DOM after mustache template processing, which was causing 2 unnecessary iterations over full data set (first through jquery selector, than through for-loop). Refactoring it inside mustache template helps saving 80-90ms on large data sets (10k+ rows)
Refactoring incomplete requests filter behavior due to inefficency in DOM manipulations. We were traversing DOM 2 more times just to hide columns that we could have avoided rendering in mustache. Factoring this logic in mustache template (`showCompletedColumn`) saves 70-80ms on 10k+ rows.
…essing Detaching history table wrapper from document before parsing it with DataTables plugin and reattaching back right after plugin has processed nested DOM. This allows to avoid huge amount of browser repaints and reflows, reducing initial page load time in Chrome from 15s to 4s for 20k+ rows
|
Can one of the admins verify this patch? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What changes were proposed in this pull request?
As described in JIRA ticket, History page is taking ~1min to load for cases when amount of jobs is 10k+.
Most of the time is currently being spent on DOM manipulations and all additional costs implied by this (browser repaints and reflows).
PR's goal is not to change any behavior but to optimize time of History UI rendering:
The most costly operation is setting
innerHTMLfordurationcolumn within a loop, which is extremely unperformant. Refactoring this helped to get time down to 10-15sSecond big gain bringing page load time down to 4s was was achieved by detaching table's DOM before parsing it with DataTables jQuery plugin.
Another chunk of improvements (1, 2, 3) was focused on removing unnecessary DOM manipulations that in total contributed ~250ms to page load time.
How was this patch tested?
Tested by existing Selenium tests in
org.apache.spark.deploy.history.HistoryServerSuite. Version of HtmlUnitDriver had a bug that was preventing rendering the full table and making testajax rendered relative links are prefixed with uiRoot (spark.ui.proxyBase)constantly fail.Changes were also tested on Criteo's spark-2.1 fork with 20k+ number of rows in the table, reducing load time to 4s.