check-new-release-gtk crashed with IOError in _buildMetaReleaseFile()
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
update-manager (Ubuntu) |
Confirmed
|
Medium
|
Unassigned |
Bug Description
Binary package hint: update-manager
This crash occurred on a system in which my home directory is served out of AFS, at a time when my Kerberos authentication had expired, and thus all of ~ was inaccessible with "Permission denied" errors.
ProblemType: Crash
DistroRelease: Ubuntu 10.10
Package: update-manager 1:0.142.20
ProcVersionSign
Uname: Linux 2.6.35-23-generic x86_64
NonfreeKernelMo
Architecture: amd64
Date: Sun Dec 5 13:22:30 2010
ExecutablePath: /usr/lib/
InterpreterPath: /usr/bin/python2.6
PackageArchitec
ProcCmdline: /usr/bin/python2.6 /usr/lib/
ProcEnviron:
PATH=(custom, no user)
LANG=en_US.UTF-8
SHELL=/bin/bash
PythonArgs: ['/usr/
SourcePackage: update-manager
Title: check-new-
UserGroups: adm audio cdrom dialout floppy plugdev video
tags: | removed: need-duplicate-check |
Changed in update-manager (Ubuntu): | |
importance: | Undecided → Medium |
Okay, it's pretty clear what's going on: update-manager fails to open the meta-release file, so it presumes it's being run for the first time and attempts to create ~/.cache/ update- manager- core, which it can't. Could the program fall back to a system-wide file (e.g. under /etc) in a case like this?
(It seems like u-m would also have this problem if the ~/.cache directory doesn't exist for whatever reason, since os.mkdir() doesn't behave like "mkdir -p"...)