IZ82: add test for Gridengine IZ 1866 (bdb server timeout)

[joga]

[Imported from gridengine issuezilla ​http://gridengine.sunsource.net/issues/show_bug.cgi?id=82]

        Issue #:      82              Platform:     All           Reporter: joga (joga)
       Component:     testsuite          OS:        All
     Subcomponent:    tests           Version:      current          CC:    None defined
        Status:       NEW             Priority:     P3
      Resolution:                    Issue type:    TASK
                                  Target milestone: milestone 1
      Assigned to:    issues@testsuite
      QA Contact:     joga
          URL:
       * Summary:     add test for Gridengine IZ 1866 (bdb server timeout)
   Status whiteboard:
      Attachments:

     Issue 82 blocks:
   Votes for issue 82:     Vote for this issue


   Opened: Wed Nov 22 03:48:00 -0700 2006 
------------------------


This was IZ 1867 in project gridengine.

Create a new testsuite test that will verify IZ 1866.

The test has to keep a cluster idle until the sge_qmaster session times out in
berkeley_db_svc.

We currently don't know the default timeout, but it is in the order of days.

Waiting several days in testsuite certainly doesn't make sense.

But it should be possible to set the timeout value in DB_CONFIG configuration
file in the database directory.
When we get the necessary information from Sleepycat (parameter name, value
units etc.), it should be possible to
- shutdown sge_qmaster
- shutdown the RPC server
- create a DB_CONFIG file and setting the timeout parameter to a short value
(e.g. 5 minutes)
- restart the RPC server
- restart sge_qmaster
- wait > 5 minutes
- submit a job. Job submission shall succeed. In buggy SGE version, qsub will
fail with an error message containing the DB_NOSERVER_ID keyword.

------- Additional comments from joga Tue Nov 8 08:17:09 -0800 2005 -------

Update from Sleepycat:
It is not possible to set the timeout parameter in DB_CONFIG.

So the only way to set the timeout parameter is via C-Interface
(DB_ENV->set_rpc_server).

We would have to introduce an environment variable containing the timeout value.
If this environmentvariable is set, the spooling framework for Berkeley DB
spooling has to pass the value in its call to DB_ENV->set_rpc_server.

Then the test would be:
- shutdown qmaster
- restart qmaster with the env variable set
- wait > timeout
- submit a job. Job submission shall succeed. In buggy SGE version, qsub will
fail with an error message containing the DB_NOSERVER_ID keyword.

------- Additional comments from joga Fri Nov 11 05:13:55 -0800 2005 -------

berkeley_db_svc has an option -I to set the timeout parameter.
So the test can in fact be written as described above.