| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| ... | |
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When dracut finds 'cdrom' in a kickstart it sets inst.repo=cdrom
yuminstall.py needs to treat this as a special case meaning that
it should use the first cdrom found, either the one mounted as the
install media, or by scanning attached devices.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
As of beta4, we shouldn't need to generate a device map for mdraid and
such:
<phcoder> a) You create entries only for IMSM, dmraid and real disks
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
When an invalid keyboard type is passed in a kickstart display an error
message that makes it clear that the value is incorrect.
|
| | | | |\
| | | | |
| | | | |
| | | | | |
into f17-branch
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
There are two things the method was doing wrong:
1. not avoiding the names of lvs in the dialog's temporary vg
2. not using the full device name when checking numbered lv names
(lv device names are of the form "%s-%s" % (vgname, lvname))
There is another change here as well: if no distinguishing info is
passed in (eg: mountpoint, swap), default to numbered names from
the start. For the first three lvs you create, instead of lv, lv00,
lv01 you'll get lv00, lv01, lv02. Consistency.
(cherry picked from commit f9df90561901f5b0f99d801e99daf78bd98cd233)
|
| | | | | |
| | | | |
| | | | |
| | | | | |
(cherry picked from commit a5c4397ff0b354a1c2243bccddd35627c7b3d9b4)
|
| | | | | | |
|
| | | | |/
| | | |
| | | |
| | | |
| | | |
| | | | |
Now that we are running from the media instead of RAM we can
lower our minimum requirements a bit. Tested with a DVD desktop
install in 512M and it worked fine.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Imported from https://github.com/cosimoc/fedora-welcome
This is for Fedora 17 only. It will be moved into its own package
for Fedora 18.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Instead of us knowing about crashkernel stuff in our code, let the
crashkernel guys own that policy.
rhcrashkernel-param comes from the following url, though I'm told
they're updating it not to end in .sh there:
http://pkgs.fedoraproject.org/gitweb/?p=kexec-tools.git;a=commitdiff;h=ca94c80a91400e9ce95bb56023ca4391d5b41d85
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
This is step one of two - step two is removing the file from the
package.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This is really just placeholder code, since I think this is going to go
away or at least completely change when newui hits. Still should something
hold that up, this'll be fixed.
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | | |
Only a few specific items in data need to be updated.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
When dracut finds 'cdrom' in a kickstart it sets inst.repo=cdrom
yuminstall.py needs to treat this as a special case meaning that
it should use the first cdrom found, either the one mounted as the
install media, or by scanning attached devices.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
As of beta4, we shouldn't need to generate a device map for mdraid and
such:
<phcoder> a) You create entries only for IMSM, dmraid and real disks
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
When an invalid keyboard type is passed in a kickstart display an error
message that makes it clear that the value is incorrect.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Now that we are running from the media instead of RAM we can
lower our minimum requirements a bit. Tested with a DVD desktop
install in 512M and it worked fine.
|
| | | |\| |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Basically the same deal as commit 45363385 - look at /run/install/repo
(where the CD will be mounted) when looking for media.
Obviously, scanForMedia should handle already-mounted media instead, but
that's something for F18.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If our methodstr is 'cdrom' then anaconda expects the CD to be mounted
at /mnt/install/source. /mnt/install is a symlink to /run/install, so
making /run/install/source a symlink to the actual mountpoint makes
anaconda able to find the CDROM.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This can actually cause some problems because "getarg nameserver" will
return "1" for the argument "nameserver=", which means you end up with
DNS1=1 in your ifcfg, which is obviously wrong.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The replacement hack for splitsep wasn't detecting the case where the
repo is just "cdrom" - no ":PATH" at the end - and trying to remove the
"cdrom:" from the front of "cdrom", which doesn't work.
In that case, we'll end up with $rest = $repo, so check for that, and
set rest="".
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
parse-kickstart needs to check the device to see if it's a MAC and do
the right thing when it is.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If there is no treeinfo the stage2 variable may still be set,
check if treeinfo or stage2 are empty and fall back to
LiveOS/squashfs.img
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
So, loader used to look for updates.img/product.img/RHUpdates under
whatever path it was given for the stage2 image. We need product.img for
variant installers (like the RHEL installer) so we need to look for
these things.
Add the anaconda_auto_updates() function to look for these things and
put them in the appropriate places. Also add unpack_updates_img() as a
helper for cleanly unpacking updates images.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Since we've removed all UI from initramfs, plain 'updates' no longer
works if you want to load an updates image from disk.
This adds support for fetching updates images, much the same way as we
do with kickstarts.
Note that I'd really like to merge all the *-genrules.sh / fetch-*-disk
junk, but time is short..
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
copytree() is a function for copying/merging directory trees - useful
for stuff like updates images.
dev_is_mounted is confusingly similar to dracut-lib's 'ismounted', so
it's been renamed for clarity.
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
mactel-boot-setup needs to be run after install in order to ensure that
Mac-specific bootloader configuration occurs. Make sure this happens.
Bug 816288.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Any non-root device that appears in /etc/fstab and requires activation
(eg: lvm, md) must be active prior to instantiating an FS instance
for an existing filesystem on that device due to current/minimum size
calculations triggered by the FS constructor.
(cherry picked from commit a8464f82866a19f33443082d9a0710036cc0a6db)
|
| | | | |
| | | |
| | | |
| | | | |
(cherry picked from commit 41f9a52d2ce0ab58535f6245859c165358030999)
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
The users file should not be og+rx, and the data it's using should be
exported.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The existing logic forces text-mode if /usr/bin/Xorg is missing. Except
it doesn't check if you're using VNC first, which caused problems on
certain POWER systems where we're not including Xorg anymore.
We could just add 'and not flags.usevnc' to the if statement but it's
already really, *really* hard to understand what's going on there.
So: this patch cleans up the logic in a way that should be equivalent,
*and* fixes the bug.
A couple notes, if you're checking my work:
- The flags.livecdInstall check was to avoid starting X during liveinst
(assuming it'd already be running on a live system), so the intent
there was:
flags.livecdInstall == flags.preexisting_x11
- The 'isS390' check in the first block was added in commit 4d42471, way
back in 2004: "enable GUI/VNC installs on mainframe again, there's no
Xorg binary". So the intent there was:
isS390 == (flags.preexisting_x11 or flags.usevnc)
Knowing that, you can replace the conditions and do a little juggling
and come up with what I have here. Or something like it.
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
loader used to set 'headless' unconditionally for s390 systems, so do
that here.
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
It appears that Lenovo systems aren't the only ones that have problems
booting from GPT labeled disks. This patch does the following:
* default to msdos labeled disks
* remove the GPT blacklist
* remove nogpt cmdline argument
* add gpt cmdline argument to force use of GPT disklabel
|
| | | | |\
| | | | |
| | | | |
| | | | | |
into f17-branch
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
BootLoader.writeKS writes an incorrect value in the kickstart file.
Fix is written by dlehman and tested on PPC.
|
| | | | |/ |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
An iso that had been dd'd to a USB stick needs to find the
packages so look at the root of the media (which dracut mounts
on /run/initramfs/live/) and bind mount that to /mnt/install/source
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
There are two things the method was doing wrong:
1. not avoiding the names of lvs in the dialog's temporary vg
2. not using the full device name when checking numbered lv names
(lv device names are of the form "%s-%s" % (vgname, lvname))
There is another change here as well: if no distinguishing info is
passed in (eg: mountpoint, swap), default to numbered names from
the start. For the first three lvs you create, instead of lv, lv00,
lv01 you'll get lv00, lv01, lv02. Consistency.
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If there is no treeinfo the stage2 variable may still be set,
check if treeinfo or stage2 are empty and fall back to
LiveOS/squashfs.img
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Patch for duped #808642.
Related: rhbz#735857
|
| | | | |
| | | |
| | | |
| | | | |
Resolves: rhbz#735857
|
