Skip to Content

Explorer connected to HANA is fast becoming a very popular platform for deploying an accelerated data discovery solution. There is still a great deal of power harnessed with BWA, a stable and mature solution accelerating BW models and agnostic data for Explorer, but the newest in-memory solution with HANA is where most of the attention is given. For this reason I wanted to address some common issues with this setup that I have seen and hopefully answers some questions. I will try to update this blog if I come across new items.

1.Making consistent HANA connections.

In the past, the way connections were specified is via CMC parameters, in CMC -> Applications -> Explorer. This is an older way of specifying connections ( Explorer 4.0 SP02 and older) but still valid. My recommendation is, do not use this method, the preferred way is to specify JDBC HANA connections via Information Design Tool (IDT). One major advantage is that these connections can survive migration to other environments using Promotion Management.

2. Connections show broken (invalid) under Manage Spaces.

There are many reasons for this. The Explorer master server is responsible in generating the data source list which includes HANA connections. When using Internet Explorer if you hover the mouse over the connection sometimes it will specify the reason. Normally a message as such “NewDB server does not respond” is one example. Some things to check for:

  1. The HANA host is not accessible from the Explorer host. Check network connectivity and port blockage. You may have to add the IP, short name and long name of the HANA server in the hosts file. Try using HANA IP instead of hostname in the connection.
  2. The user specified in the connection may not have rights to access any of the Analytic or Calculation Views. To test this, use HANA Studio and with the same account right click on the View then select Data Preview to see if you are able to see the data.
  3. Delete any HANA connections, that are invalid and not needed. If they are needed validate them against a) and b) above to make them valid. I have seen many instances where the master server will flag good connections broken if there are others which are not valid.

Note: To save time in checking connection validity in Explorer,  you don’t have to log off and log back in,  just close Manage Spaces tab and re-open it, master server will auto rescan the data sources list for changes.

3. It takes a long time for Manage Spaces list to show up.

I have added this section because invalid HANA connections will cause considerable delay in the Manage Spaces list display. By delay meaning over one minute. In reality this should normally take only a few seconds. Invalid HANA connections is one cause for this delay because master server is waiting for the invalid connection to time out. I can say that there is a code change coming in Explorer 4.1 SP03 to handle this specific timeout differently. One exception to the statement that it should take seconds, is that there will be additional delay introduced in instances where there is a really large number of folders in the system (>50K). Master server scans each one looking for Excel files this being another data source option in the list.

4. How to check the list of views that should appear under a connection.

The master server sends out the query below in order to get the list of views available for a specific connection and visible to the user used in the connection. This can be ran in HANA Studio to cross check using same user then check the list generated:

mdx select “COLUMN_OBJECT”, “CATALOG_NAME”, “CUBE_NAME”, “DESCRIPTION” from BIMC_CUBES

5. Query performance checks.

When exploring (loading) an information space there are two queries dispatched to HANA. One for the facets (top section) and one for the visualization (Chart and table section). The facets query will take the most time because it has to aggregate the measure for each facet. To capture the queries, in CMC enable logging to High for the Exploration server and load the information space. In the log file “~\logging\explorerExploration…log” search for string “Execute SQL” and the line will show the select statement dispatched to HANA. The longer query is the one for the facets. Run this query directly in HANA Studio to get an idea of running time and what to expect. Explorer will add a bit of overhead but it should not take much longer than what it takes in Studio. One feature was added to bypass running the facets query till later as far back as Explorer 4. SP06. The option to not load the facets, was introduced specifically for this purpose, a selection available when creating the information space.

do not load facet.png

To report this post you need to login first.

8 Comments

You must be Logged on to comment or reply to a post.

  1. Mike Howles

    We are running into your scenario in item 3 — We are getting timeouts when trying to expand Excel Spreadsheets when under Manage Spaces.  Any options besides increasing the request.timeout parameter?

    (0) 
    1. George Pertea Post author

      Hi Michael,

      do you have a lot of Excel files and folders in your sysem?

      I think this will give you the number:

      SELECT si_name  FROM CI_INFOOBJECTS  WHERE SI_KIND = ‘Excel’ and SI_INSTANCE_OBJECT=0

      Normally the list agregation happens when you click Manage Spaces, I have not heard any complains when you actually expand the Excel tree.

      If you increase the request.timeout (say 360  or -1) do you eventually get the list?

      George

      (0) 
      1. Mike Howles

        George,

        Thanks for the response.  We tried bumping up request.timeout actually all the way to 15 minutes (900 seconds) and it never comes back…  Also, I’m seeing this in the logs:

        No idea if it’s related, but the entry in the log happens as soon as I click on Manage Spaces (not after the timeout):

        at com.businessobjects.datadiscovery.admin.commands.UniverseTree$DataSourceTreeBuilder.run(UniverseTree.java:535)

        |CABBC94FB285439298AA13AE67A1CE251e|2014 05 02 16:33:10.068|-0500| |Error|  | | |explorerMaster_BI4PROCP1.ExplorerMasterServer| 4900|13703|pool-45-thread-1| ||||||||||||||||||||com.businessobjects.datadiscovery.admin.commands.UniverseTree$DataSourceTreeBuilder||Error occured while building the data source tree for driver: com.businessobjects.datadiscovery.dataprovider.newdb

        1. java.lang.NullPointerException: while trying to invoke the method java.util.List.isEmpty() of a null object loaded from local variable ‘dsTreeItems’

        at com.businessobjects.datadiscovery.admin.commands.UniverseTree$DataSourceTreeBuilder.run(UniverseTree.java:519)

        at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:895)

        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:918)

        at java.lang.Thread.run(Thread.java:743)

        (0) 
      2. Mike Howles

        Also, results from this query:

        SELECT TOP 999999 si_name  FROM CI_INFOOBJECTS  WHERE SI_KIND = ‘Excel’ and SI_INSTANCE_OBJECT=0

        2053 objects came back.

        (0) 
        1. George Pertea Post author

          Hi Michael,
          That’s a few Excel files but not too  bad.

          The error is related to HANA connections. Do you have any HANA connections specified in IDT or BWA/HANA connections in CMC-Applications-Explorer?

          What Explorer patch level are you on and is this a clustered environment or all on same host (tomcat/cms/explorer)?

          Also since you set the timeout to 900 seconds it will be ignored. Set it to -1 and restart Tomcat and see how long it takes to show the list.

          Can you send me the Master server trace and a Fiddler trace after you set the timeout to -1.

          (0) 
          1. Mike Howles

            Thanks for the info, George.  So “newdb” references I can ignore as we are not a HANA shop?  (Just confirming)

            We have BWA but are not using it yet on this server, however I understand anything with “trex” would be BWA, so I’m ignoring those.

            We are on SBOP and Explorer 4.1 SP2 Patch 3.  I’ll work on a Fiddler trace and Master trace (any certain level to set it at?)

            (0) 
            1. George Pertea Post author

              Hi Michael,

              yes newdb references are for HANA. Also let me know if you have any entries of any kind in CMC-Applications-Explorer-Properties

              Set tracing to High and also start Fiddler, then log in Explorer and go through the workflow to show your list.

              Send both logs to me george.pertea@sap.com.

              Thanks,
              George

              (0) 

Leave a Reply