#2913: Memory Leak on SVN Repo Access using tracd or WSGI
---------------------------------------+------------------------------------
Reporter: edsuom | Owner: jonas
Type: defect | Status: new
Priority: high | Milestone: 0.10
Component: browser | Version: devel
Severity: major | Resolution:
Keywords: wsgi, tracd, memory, leak |
---------------------------------------+------------------------------------
Comment (by cboos):
I have some big reservations about that patch.
As I said before, the reality of the leak itself is still
questionable for me: I simply don't have it.
If you follow the instructions in TracSubversion,
in particular if you pay attention to the compatible SWIG versions,
I don't think a leak of pool objects will happen.
Do someone else also have the leak? mgood?
What happens when you insert the following debug statements?
{{{
#!diff
Index: svn_fs.py
===================================================================
--- trac/versioncontrol/svn_fs.py (revision 3050)
+++ trac/versioncontrol/svn_fs.py (working copy)
@@ -143,6 +143,11 @@
# Destroy pool
self.apr_pool_destroy(self._pool)
+ import gc
+ gc.set_debug(gc.DEBUG_UNCOLLECTABLE)
+ print "GarbageCollect: ", gc.collect()
+ print "Garbage: ", len(gc.garbage)
+
# Clear application pool and terminate APR if necessary
if not self._parent_pool:
application_pool = None
}}}
You should see this kind of output in your log:
{{{
...
GarbageCollect: 5
Garbage: 0
Trac[svn_fs] DEBUG: Closing subversion file-system at
C:/Workspace/local/svn/trac-svnrepos
GarbageCollect: 0
Garbage: 0
}}}
--
Ticket URL: <http://projects.edgewall.com/trac/ticket/2913>
The Trac Project <http://trac.edgewall.com/>
_______________________________________________
Trac-Tickets mailing list
[email protected]
http://lists.edgewall.com/mailman/listinfo/trac-tickets
