hadoop-hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Edward Capriolo (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HIVE-833) Web Interface new features
Date Wed, 16 Sep 2009 01:10:57 GMT

    [ https://issues.apache.org/jira/browse/HIVE-833?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12755828#action_12755828

Edward Capriolo commented on HIVE-833:

I thought I would list the current features for the talk.

Schema Browsing

  Uses HiveMetaStoreClient .getDatabases()
  iterates the databases and produces links to show_database.jsp? with the name encoded

  Uses HiveMetaStoreClient .getDatabase("default") (hard code should be the param) OOPSE
  calls getTables()
  iterates tables producing a link to show tables with db and table params

  Uses HiveMetaStoreClient .getTable(db,table)
  gets StorageDescriptor for a table()
  shows information about the table bucketCols,FieldSchema,sortOrder, inputOutputFormat, etc.


  iterates System.getProperties()
  iterates System.getEnv()
   This uses the static ExecDriver.runningJobKillURIs.keySet() to list jobs and their kill
   (the Kill URI here could be a link)

Session Stuff


  A loop-around page that sets HWIAuth as a session key
  HWIAuth maybe can be replaced with the common hadoop UGI object if it supports equals()
and comparable() ??

    this takes a string a session name and creates a session if the name is not in use
    forwards to session_manage.jsp

    setting up the query is done from here there is a form to set the query,silent mode, and
an activate box
    depending on the stae of your query changes may or may not be acceptable

    iterate values in hiveconf object



    a primative file chunk viewer that can view result files

Future Nodes
So most features are implemented in separate JSP pages. This is clean and relatively easy
to follow
but we can condense this with wiki and have a really smooth UI.

These are my early suggestions
left Side could be an Ajax style schema browser. We can collapse Database and table information.
This would not take up that much space and would be similar to the way many tools are designed.

Session_diagnostics and Diagnostics. These should be on the right side in an ajax syle window.
Really we don't need to show all values at once just provide up and down arrows. So these
could be collapsed by default

Top is query box. Depending on the session state we can enable/disable the form elements.
This could be on an ajax update timer. We should also show the session_status READY, QUERY_RUNNING,
Middle Box is ResultBucket we can update this on interval and would want to dedicate a certain
number of lines for it
Middle Box is OutFile chunk viewer browser (if we are dumping to a local outfile)
Middle Box is ErrFile chunk viewer browser (if we are dumping to a locel errfile)
Bottom box is a box with 5 lines or so dedicated to hive_history. This would be collapsible
and sizable

(You never thought you could mock up ajax with ascii. Did you?)
| Login    | session|ecapriolo                   |status|ready| Diagnostics|
| ecapriolo|   Query Box				      |  hive_home | 
| ecapriolo|   ---------------------------------------------  | Session Dia| 
|----------|   |show tables;				   |  |	 dfs.name  |		
| schema   |   |					   |  |	---------- |
|----------|   --------------------------------------------|  | Saved Queri| 
|db1       |   | Run | Save |				      |  my query  |	
|-table1   | _______________________________________________  |  my query2 |
|--col1    | | Session Bucket      export XML,CSV  Refresh5|  |	           |
|--col2    | |  X    |   Y     |    z			   |  |            |
|          | |  5    |   hi    |    5			   |  |            |
|          | ----------------------------------------------|  |		   |
|          | |Hive Hist Bucket    export XML,CSV  Refresh5 |  |	           |
|          | |  X    |   Y     |    z			   |  |            |
|          | |  5    |   hi    |    5			   |  |            |
|          | ----------------------------------------------|  |		   |

> Web Interface new features
> --------------------------
>                 Key: HIVE-833
>                 URL: https://issues.apache.org/jira/browse/HIVE-833
>             Project: Hadoop Hive
>          Issue Type: Improvement
>          Components: Web UI
>            Reporter: Edward Capriolo
>            Assignee: Edward Capriolo
> These are updates we need to discuss.
> * build file clean up
> * implement wikit
> * make launch target for debugging ease
> * standard features where you can "build and save" queries for reuse
> * queries and retrieve results in CSV/XLS format
> * replace JSP with wiki classes
> * create better JUnit Tests
> * provide a REST/XMLRPC capability
> * provide better interface (ajax?)
> * easy to plug 
> * HiveServer as backend
> Also we can attach some mock up of what the features should look like. 
> After we discuss these we can break them out into separate Jira Issues.

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message