Opened 10 years ago

Last modified 6 years ago

#448 new defect

IZ2347: NFS problems with java aimk build

Reported by: crei Owned by:
Priority: high Milestone:
Component: sge Version: current
Severity: Keywords: build
Cc:

Description

[Imported from gridengine issuezilla http://gridengine.sunsource.net/issues/show_bug.cgi?id=2347]

        Issue #:      2347             Platform:     All       Reporter: crei (crei)
       Component:     gridengine          OS:        All
     Subcomponent:    build            Version:      current      CC:    None defined
        Status:       NEW              Priority:     P2
      Resolution:                     Issue type:    DEFECT
                                   Target milestone: ---
      Assigned to:    andreas (andreas)
      QA Contact:     andreas
          URL:
       * Summary:     NFS problems with java aimk build
   Status whiteboard:
      Attachments:

     Issue 2347 blocks:
   Votes for issue 2347:


   Opened: Thu Aug 16 08:22:00 -0700 2007 
------------------------


The testsuite builds often breaks because of some build
problems for aimk java compile tasks using NFS. This happens
when the java build host is not the local host.

The ant call <delete dir="DIR_TO_DELETE"/> sometimes
makes problems on NFS directories, because call reports
errors when there are .nfsXXX files in the directory.

A workaround by using the call
<delete dir="DIR_TO_DELETE" failonerror="false"/>
which does not report an error is not a good solution
because the deletion might be done after new files
are created in the directory.

A solution might be:
o Implement a new ant target for deleting directories and
  use it instead of <delete dir>

This would centralize the problem to one target call!

Possible implementations:

a) move the specified dir to a temporary filename
b) delete the moved directory
=> not sure if this works without errors

or

a) delete directory with failonerror=false
b) wait for removal of directory with a timeout of 70 seconds
   (NFS needs sometimes 60 seconds to distribute changes)
c) after timeout: delete directory with failonerror=true

or

a) add a property for NFS hostname and connection method
b) delete the directory on the NFS server
c) wait that the directory is deleted and not available on
   the NFS client before creating new files in the directory

or

some other solutions welcome!!!

Change History (0)

Note: See TracTickets for help on using tickets.