[jira] [Commented] (HBASE-25641) Improve storeFile.jsp performance on really large files

["Anoop Sam John (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HBASE-25641?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17298533#comment-17298533 ] 

Anoop Sam John commented on HBASE-25641:
----------------------------------------

Seems the FileInfo will not have all the info what this page displays. Like info abt value sizes and even key length different percentile values and we end up reading entire file !
HFilePrettyPrinter#processFile opens the HFile reader with CacheConfig.DISABLED means, even if this file is present in cache, we will not use that cache at all.  It is not just this read will avoid caching blocks to BC if not found.  The reads wont make use of cache at all.  May be we need to revisit that also in HFilePrettyPrinter?  It can make use of the cache if data is present (I know its not completely related to this jira. still asking. Need another jira)

We might have to add more details about key, value, row lengths to FileInfo  while writing HFile?  Display only this FileInfo in UI. (If older files dont have all these info, lets not show those)?

> Improve storeFile.jsp performance on really large files
> -------------------------------------------------------
>
>                 Key: HBASE-25641
>                 URL: https://issues.apache.org/jira/browse/HBASE-25641
>             Project: HBase
>          Issue Type: Improvement
>          Components: regionserver, UI
>    Affects Versions: 2.3.0
>            Reporter: Nick Dimiduk
>            Priority: Minor
>
> Since HBASE-7171, we have a nice UI for clicking down into the details of an hfile on a regions server. When the store file is really large, the performance of this page is quite bad. It can take several minutes to summarize a 30-50gb hfile. We can do better.



--
This message was sent by Atlassian Jira
(v8.3.4#803005)