DESCRIPTION

Try to recover after deleting opened files on NFS

Fixes various LocalRepositoryTestCase based tests, which were failing if
executed on NFS mount, see Bug 440182. The current workaround doesn't
really checks for NFS (have no idea how to do this in Java) but allows
tests running on "/home" to run "delete" cleanup twice (second "delete"
round deletes unexpected nfs zombie files). The assumption is that this
workaround should at least help *nix users with their NFS mounted home
directories.

Ideally the tests should be fixed to properly cleanup opened .pack file
handles, but I've spent one day hunting the root cause on one single
test without success, due the asynchronous nature of the resources
creation/cleanup.

Bug: 440182
Change-Id: I5e879daed954aa35de197d8e1ea43c9fbccca7d4
Signed-off-by: Jayvion Booth xxx@xxx.xxx

COMMENTS

author: Jayvion Booth
date: 2014-07-31 08:55:27.000000000

Uploaded patch set 3.

-------------------------------------
author: Brenden Conley
date: 2014-07-31 09:19:29.000000000

Patch Set 3:

Build Started https://hudson.eclipse.org/egit/job/egit.gerrit/6140/

-------------------------------------
author: Brenden Conley
date: 2014-07-31 09:33:04.000000000

Patch Set 3: Verified+1

Build Successful 

https://hudson.eclipse.org/egit/job/egit.gerrit/6140/ : SUCCESS

-------------------------------------
author: Juliet Cantu
date: 2014-08-02 08:18:50.000000000

Patch Set 3: Code-Review-1

(1 comment)

Line:21, /COMMIT_MSG -> I think we used to be synchronous, even i EGit...

-------------------------------------
author: Jayvion Booth
date: 2014-08-02 08:25:59.000000000

Patch Set 3:

This isn't what I observed. Don't see the code right now, but there was cleanup of resources in Jgit done by spawning  new threads. I tried to debug it but as said, the timing diring debug session is different so that I never was able to understand who and where must properly wait for cleanup.   The proposed workaround helps me really, because without it tests fail almost always on NFS.

-------------------------------------
author: Jayvion Booth
date: 2014-08-02 08:37:49.000000000

Patch Set 3:

BTW I always saw few opened .pack files from *different* repositories used during tests on Linux just before the test failed. The problem there is that repository.close() don't close opened object databases because for some reason the internal use count says that nothing is opened (however the pack files were opened!).

-------------------------------------
author: Jayvion Booth
date: 2014-09-24 12:35:39.000000000

Patch Set 4: Patch Set 3 was rebased

-------------------------------------
author: Gerrit Johnathan Pham
date: 2016-07-13 17:11:03.000000000

Change has been successfully merged into the git repository by Jayvion Booth

-------------------------------------
