Jan Pazdziora wrote:
On Thu, Jan 08, 2009 at 09:44:08AM -0500, Pradeep Kilambi wrote:
As I recall cliff mentioning to me once.. there is a bug in the
specspo libraries with rpm which cause apache heap corruption due to
specspo translating strings within memory when you use rhnpush to
upload packages into a Satellite it would not free memory correctly..
I may have been fixed in specspo now..I'm not sure about that.
oops I meant "they may have fixed in specspo now" .. not I :)
I've just tried rhnpush on my Spacewalk 0.4 and the result is:
1) When specspo is installed, the size of httpd processes after I do
rhnpush is the same as when it is not installed, and it does not
seem to grow after multiple rhnpushes.
2) Even if I start the Apache as
( export LANG=cs_CZ.UTF-8 ; /etc/init.d/httpd start )
the rpm's name stored in the database (as shown in the WebUI) is
the original English one, and the size of processes is the same as
when LANG is not set (or set to en_US.UTF-8).
I think I'm going to file a bugzilla for removal of that Conflict.
Cast your votes if you want it aligned to 0.4 or 0.5 triage.
0.5 would make me feel more comfortable - just because this was such a
bad bug (as Mirek pointed to comments in an old customer driven -
non-public bugzilla). I would not like to see the seg faults within
apache reoccur and regress for future Satellite versions based on
Spacewalk. The bug was basically apache -> mod_python -> rpm python
libraries and specspo doing 'bad things' within memory for parsing the
headers of the rpms as we attempt to upload them via apache+mod_python
into Satellite - where headers go into DB and package onto disk.
Cliff.
--
Clifford Perry
Manager, Satellite Engineering
Red Hat, Inc.
http://www.redhat.com/
+1 919 754 4403
RHCA / RHCE# 805007680128201
_______________________________________________
Spacewalk-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/spacewalk-devel
