Bug 358725

Summary:	hggit: too many open files on hg pull
Product:	Gentoo Linux	Reporter:	Fabian Groffen <grobian>
Component:	New packages	Assignee:	Dirkjan Ochtman (RETIRED) <djc>
Status:	RESOLVED INVALID
Severity:	normal	CC:	grobian
Priority:	Normal
Version:	unspecified
Hardware:	All
OS:	Linux
URL:	https://github.com/schacon/hg-git/issues/176
Whiteboard:
Package list:		Runtime testing required:	---

Description Fabian Groffen gentoo-dev

2011-03-13 16:16:33 UTC

% hg pull pulling from git+ssh://git@git.overlays.gentoo.org/proj/portage.git
importing Hg objects into Git
["git-upload-pack '/proj/portage.git'"] Counting objects: 3664, done.
Compressing objects: 100% (2573/2573), done.
Total 3664 (delta 2289), reused 1743 (delta 1091)
importing Git objects into Hg
abort: Too many open files: /export/gentoo/portage-hg/.hg/git/objects/pack/pack-9fb8d5a8932a495a2ed28ac993b4329a1ad3c680.pack
%

This problem stems from commit https://github.com/schacon/hg-git/commit/59c7b6d804126bde4fb85409b1cc66c01de4132c the code block for hgrepo.py around line 50 opens tf, but doesn't close it. Adding tf.close() after the tf.read line, resolves the too many open files problem.


patch is simple and straightforward:
--- hgrepo.py.orig       2011-03-13 17:09:21.000000000 +0100
+++ hgrepo.py     2011-03-13 17:09:43.000000000 +0100
@@ -55,6 +55,7 @@
             if os.path.exists(tagfile):
                 tf = open(tagfile, 'rb')
                 tagdata = tf.read().split('\n')
+                tf.close()
                 td = [line.split(' ', 1) for line in tagdata if line]
                 return dict([(name, bin(sha)) for sha, name in td])
             return {}

already reported upstream

Comment 1 Fabian Groffen gentoo-dev

2011-03-13 16:59:38 UTC

the fix works on the platform I tested by coincidence, the real problem is in dulwich, so closing.  sorry for the noise.