Anaconda-devel November 2011

anaconda-devel@lists.fedoraproject.org

25 participants
38 discussions

[PATCH rhel5-branch] Ignore duplicate opening of installed rpm package file (#506361).
by Radek Vykydal 15 Nov '11

15 Nov '11

This is second take on the fix, the first one broke some other methods of installation (bug #752507) The patch is relative to 5.7 not to the previous take. Yum calls open file callback twice when installing some packages which makes our callback try second download and fail in specific case of nfs: method and http: additional repository. The fail is caused by self.method.unlinkFilename(fn) in RPMCALLBACK_INST_CLOSE_FILE being driven by method (i.e. do nothing for nfs, remove rpm file from cache for http), not by repo in question. In the problematic case the package of additional http repository is not removed from cache because the install method is nfs. As a consequence, in the second call of RPMCALLBACK_INST_OPEN_FILE repo.getPackage(po) fails to download the package. The second call of the OPEN_FILE callback is a bug in yum. Resolves: rhbz#506361 --- yuminstall.py | 8 ++++++-- 1 files changed, 6 insertions(+), 2 deletions(-) diff --git a/yuminstall.py b/yuminstall.py index 209a8d6..6d2b670 100644 --- a/yuminstall.py +++ b/yuminstall.py @@ -134,6 +134,10 @@ class simpleCallback: nvra = "%s" %(po,) self.instLog.write(self.modeText % (nvra,)) + if nvra in self.files: + fn = self.files[nvra].name + else: + fn = None self.instLog.flush() self.files[nvra] = None @@ -142,8 +146,8 @@ class simpleCallback: while self.files[nvra] == None: try: - fn = repo.getPackage(po) - + if not (fn and os.path.exists(fn)): + fn = repo.getPackage(po) f = open(fn, 'r') self.files[nvra] = f except NoMoreMirrorsRepoError: -- 1.7.4

3 4

[PATCH] ARCHIVE_DEFAULT_BYTES_PER_BLOCK no longer exists in libarchive-3.0.0
by Chris Lumens 14 Nov '11

14 Nov '11

--- loader/unpack.c | 3 +-- 1 files changed, 1 insertions(+), 2 deletions(-) diff --git a/loader/unpack.c b/loader/unpack.c index f7a5777..0323d11 100644 --- a/loader/unpack.c +++ b/loader/unpack.c @@ -152,8 +152,7 @@ int unpack_archive_file(char *filename, char *dest) { return rc; } - rc = archive_read_open_filename(a, filename, - ARCHIVE_DEFAULT_BYTES_PER_BLOCK); + rc = archive_read_open_filename(a, filename, 10240); if (rc != ARCHIVE_OK) { logMessage(ERROR, "error opening %s (%s:%d): %s", filename, __func__, __LINE__, -- 1.7.6

2 1

udevadm settle hangs forever and ever on liveinst
by Chuck Anderson 14 Nov '11

14 Nov '11

I just did a F16 LiveCD (USB) install on a ThinkPad T61. The installer kept hanging due to "udevadm settle" hanging. I had to repeatedly type "killall udevadm" from a terminal everytime the storage code was doing something just so the install would complete. The install completed successfully and the system boots despite that. I know I've seen this before on F15 at least, but only on this laptop. Any hints about how to debug this? Thanks.

1 0

[PATCH] Don't use the rpmdb to figure out upgrade target arch (#748119).
by Chris Lumens 10 Nov '11

10 Nov '11

This will fail if /var is on a separate partition, since we do not have the logic to mount additional partitions at this point. Instead, just run a chrooted /bin/arch which should always succeed. --- pyanaconda/storage/__init__.py | 25 +++++-------------------- 1 files changed, 5 insertions(+), 20 deletions(-) diff --git a/pyanaconda/storage/__init__.py b/pyanaconda/storage/__init__.py index 4701074..079a3b8 100644 --- a/pyanaconda/storage/__init__.py +++ b/pyanaconda/storage/__init__.py @@ -1460,28 +1460,13 @@ class Storage(object): return 0 def getReleaseString(): - relArch = None relName = None relVer = None - import rpm - iutil.resetRpmDb() - ts = rpm.TransactionSet(ROOT_PATH) - - # We get the arch from the initscripts package, but the version and name - # must come from reading the release file. try: - # pylint: disable-msg=E1101 - mi = ts.dbMatch('provides', 'initscripts') - except Exception: - # This could happen in a variety of cases, but the biggest one is we're - # examining an installed system that doesn't use RPM. Raise an - # exception for the caller to handle. - raise ValueError - - for h in mi: - relArch = h['arch'] - break + relArch = iutil.execWithCapture("arch", [], root=ROOT_PATH).strip() + except: + relArch = None filename = "%s/etc/redhat-release" % ROOT_PATH if os.access(filename, os.R_OK): @@ -1548,8 +1533,8 @@ def findExistingRootDevices(anaconda, upgradeany=False): device.teardown(recursive=True) if arch is None: - # we failed to determine the arch (for instance when there is a - # corrupted rpm database on the target system) + # we failed to determine the arch (for instance when we cant' + # run a binary on the target system) log.info("findExistingRootDevices: no arch.") continue -- 1.7.6

1 0

new UI stuff - PUSHED!
by Chris Lumens 10 Nov '11

10 Nov '11

Since I didn't hear any cries of anguish, I figured no one had any complains with the new UI stuff I'd previewed on here and just went ahead and pushed it. It's all on a new branch named "newui", strangely enough. Here's what you are going to find: - A new widgets/ directory. This is all the C code for custom GTK+ widgets, documentation, and glade integration files. There's also changes to the spec file to build anaconda-widgets and anaconda-widgets-devel as packages. If you want to work with the new UI on a purely python level, I suggest you build and install these packages. Note: widgets/ is a completely separate autoconfed subdirectory that gets configured as part of the top-level ./configure. Right now you have to pass --enable-introspection --enable-gtk-doc yourself. If anyone wants to take a look at how to make these the defaults, be my guest. - A new pyanaconda/ui tree. This is all the Python code that will eventually take the place of iw/ and textw/. The layout should be pretty self-explanatory. Right now there's only gui/ but others can follow later. I've only got placeholder stuff in here for language and upgrade. We can start replacing the placeholder stuff real soon, though. And then what's broken? - anaconda will only let you do a graphical install right now. - The install won't run very far because there's nothing past the first hub. - I need to send off some lorax patches to include yet more libraries, theming engines, and icons so GTK3 will look right. I'm either going to do that right now or tomorrow. If you try running it before I get to it, you may get crashes. Just stand by. - Exception handling likely crashes. That's because I haven't yet made python-meh aware of the new UI. This will probably come much, much later. So, I'm testing this stuff by building a new installation ISO with my updated lorax and the widgets/ stuff included and then adding the pyanaconda/ui/ and other anaconda changes as an updates.img. Due to how the g-i-r stuff fits together with its repo and the overrides, putting the widgets/ stuff into an updates.img does not yet work. We'll get there. Anyway, now we can start looking at merging in the pieces everyone else has been working on. For now, I'd appreciate it if everyone would run their commits by me first. I don't want to be a jerk about it, but I want to make sure we're all on the same page here. Also, I am not going to worry too much about cleaning up existing UI code as we go. My plan is to build the new UI completely separately and then after it's to a reasonable state, start removing old UI code. Please let me know if you have any questions either about how to test, how it all fits together, or what you can be working on. There's plenty to do. - Chris

1 0

anaconda storage rewrite ideas
by Tom Horsley 09 Nov '11

09 Nov '11

Labels: Human readable partition labels are incredibly valuable. Especially when you have many partitions with different distros installed for multi-booting. You can give the partitions meaningful labels and see the labels to help sort out an otherwise confused mass of partitions. In particular, if you have decided to discard an old release and install a new one, the labels are extremely useful for making absolutely sure you pick the right partition to reformat and install on. Anaconda seems to go out of its way to not show labels and (I've heard) I won't even be able to see them when clicking the Edit button anymore. Don't do that. Bring them back. They are absolutely the only human controlled and human readable bit of information that can be used to identify a partition. Speaking of human control, it would sure be nice if I could set the label as well when I use anaconda to format a partition (though this isn't as important since the partition with no label on it is a good way to identify the one anaconda just formatted - but if I formatted multiple partitions, even that could get tricky). see: https://bugzilla.redhat.com/show_bug.cgi?id=574121 Model numbers: See this: https://bugzilla.redhat.com/show_bug.cgi?id=574120 (which seems to have reappeared in the final fedora 16 when it wasn't in the alpha or beta). Some really geeky users may have the model numbers of their disks memorized, but mostly any screen that has no info other than disk model numbers is going to be really useless for making a choice of any kind. What do I want? I'd eliminate the multiple screens you have to advance through to make choices about what you want to do with disks. I'd replace it with a single screen that shows as much information as can be obtained about every disk, but with the disks "locked" so they can't be changed accidentally. When I decide which disks I want to clobber (which I can do much more easily with things like the partition labels being displayed), I could then unlock them and proceed to select partitions to reformat or free space to turn into new partitions.

1 0

pushing new UI stuff
by Chris Lumens 07 Nov '11

07 Nov '11

I'd like to push a "newui" branch to the git repo and add the following patch to that branch: http://clumens.fedorapeople.org/UI/0001-Add-our-custom-GTK-widgets-in-as-a-… This is a big patch that adds my custom GTK+ widgets into the anaconda source tree in a widgets/ directory, complete with a sub-configure script and two subpackages in the .spec file. So far, I've not seen any complaints raised about this approach, so it seems like a good time to push it and start integrating other people's work. The ui-demo stuff is going to remain separate for a little bit, at least until we've had time to figure out if it's the right approach or not. Comments? - Chris

1 0

Friday patch dump
by David Lehman 05 Nov '11

05 Nov '11

This is a bunch of patches that have been sitting around waiting for me to get around to sending them out. Patch 6 should really have gone onto f16-branch, but I didn't catch it until yesterday.

3 11

[pykickstart][PATCH] Add --boot-drive option to bootloader command.
by David Lehman 04 Nov '11

04 Nov '11

--- pykickstart/commands/bootloader.py | 19 +++++++++++++++++++ pykickstart/handlers/control.py | 2 +- 2 files changed, 20 insertions(+), 1 deletions(-) diff --git a/pykickstart/commands/bootloader.py b/pykickstart/commands/bootloader.py index b227fac..bb40a18 100644 --- a/pykickstart/commands/bootloader.py +++ b/pykickstart/commands/bootloader.py @@ -215,6 +215,25 @@ class F15_Bootloader(F14_Bootloader): op.add_option("--md5pass", action="callback", callback=password_cb, nargs=1, type="string") return op +class F17_Bootloader(F15_Bootloader): + def __init__(self, writePriority=10, *args, **kwargs): + F15_Bootloader.__init__(self, writePriority, *args, **kwargs) + + self.bootDrive = kwargs.get("bootDrive", "") + + def _getArgsAsStr(self): + ret = F15_Bootloader._getArgsAsStr(self) + + if self.bootDrive: + ret += " --boot-drive=%s" % self.bootDrive + + return ret + + def _getParser(self): + op = F15_Bootloader._getParser(self) + op.add_option("--boot-drive", dest="bootDrive", default="") + return op + class RHEL5_Bootloader(FC4_Bootloader): removedKeywords = FC4_Bootloader.removedKeywords removedAttrs = FC4_Bootloader.removedAttrs diff --git a/pykickstart/handlers/control.py b/pykickstart/handlers/control.py index 0a0c494..ed77b4f 100644 --- a/pykickstart/handlers/control.py +++ b/pykickstart/handlers/control.py @@ -769,7 +769,7 @@ commandMap = { "authconfig": authconfig.FC3_Authconfig, "autopart": autopart.F16_AutoPart, "autostep": autostep.FC3_AutoStep, - "bootloader": bootloader.F15_Bootloader, + "bootloader": bootloader.F17_Bootloader, "cdrom": method.F14_Method, "clearpart": clearpart.FC3_ClearPart, "cmdline": displaymode.FC3_DisplayMode, -- 1.7.3.4

2 2

[firstboot] Disable minimize and maximize buttons on s-c-u (#747829)
by Martin Gracik 04 Nov '11

04 Nov '11

--- firstboot.spec | 3 ++- modules/create_user.py | 19 +++---------------- 2 files changed, 5 insertions(+), 17 deletions(-) diff --git a/firstboot.spec b/firstboot.spec index 8e950a0..93e9a18 100644 --- a/firstboot.spec +++ b/firstboot.spec @@ -18,7 +18,8 @@ BuildRequires: gettext BuildRequires: python-devel, python-setuptools-devel BuildRequires: systemd-units Requires: pygtk2, python -Requires: setuptool, libuser-python, system-config-users, system-config-date +Requires: setuptool, libuser-python, system-config-date +Requires: system-config-users >= 1.2.111-1 Requires: authconfig-gtk, python-meh Requires: system-config-keyboard Requires: python-ethtool diff --git a/modules/create_user.py b/modules/create_user.py index 5408f67..3050852 100644 --- a/modules/create_user.py +++ b/modules/create_user.py @@ -39,6 +39,7 @@ _ = lambda x: gettext.ldgettext("firstboot", x) N_ = lambda x: x sys.path.append("/usr/share/system-config-users") +import mainWindow import userGroupCheck class moduleClass(Module): @@ -512,19 +513,5 @@ class moduleClass(Module): confirmIcon.hide() def _runSCU(self, *args): - i = gtk.Invisible() - i.grab_add() - - pid = start_process("/usr/bin/system-config-users") - - while True: - while gtk.events_pending(): - gtk.main_iteration_do() - - child_pid, status = os.waitpid(pid, os.WNOHANG) - if child_pid == pid: - break - else: - time.sleep(0.1) - - i.grab_remove () + reload(mainWindow) + win = mainWindow.mainWindow(firstboot=True) -- 1.7.5.4

1 0

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

Anaconda-devel November 2011