summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
...
| | * | | New version.Brian C. Lane2012-02-203-286/+331
| | | | |
| | * | | use a dracut shutdown hook to eject media (#787461)Brian C. Lane2012-02-203-32/+15
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Now that we are running from the squashfs.img on the install media we cannot eject it until shutdown. Anacona now uses a dracut shutdown hook to handle ejecting any media that needs to be ejected.
| | * | | add dracut shutdown eject hook function (#787461)Brian C. Lane2012-02-202-0/+26
| | | | | | | | | | | | | | | | | | | | dracut runs these from the initramfs just before shutdown/reboot
| | * | | The createSuggested methods have changed name (#791204, #795058).Chris Lumens2012-02-202-4/+6
| | | | |
| | * | | iscsi: no discovery on each target login (#752066)Radek Vykydal2012-02-203-61/+115
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Don't redo discovery each time when logging to a node. Skip it if already logged to a node on the same address. Instead, store the node list on the first discovery. Next discovery would rewrite information about credentials used for previous login stored by iscsilib's setAuth() (and accessed later by getAuth()) in database in /var/lib/iscsi/nodes. This is port of following commits from rhel6-branch: commit cc8cfdde3fb7bae7152075504674f6d0a7195fe4 commit dda2355ca424eb97fbca0a144260c9e9cb151a3a The patches also consolidate code in iscsi.py (UI and ks case). Similar consolidation has been already done on master with commit 06ffb569eb417263b2936dfe97969e3d36653e39, but it went a bit further (removal of addTarget function and moving its code to kickstart.py) with outlook of removing some limits of kickstart iscsi setup. I am keeping only the consolidation from rhel6-branch which goes just to the present functionality. I prefer to have the same code for rhel6-branch and master and to do the next step - moving what remained of addTarget after consolidation into kickstart so for finer ks command options - when the options will actually be added (which would probably happen on rhel6-branch first anyway).
| | * | | Use libpwquality to check root password strength (#755883)Martin Gracik2012-02-203-9/+11
| | | | |
| | * | | Generate repo= ks command only for repos added by user (#738577)Radek Vykydal2012-02-201-0/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Don't generate it for base repo which corresponds to kickstart install url/nfs/cdrom command. Don't generate if for addon repos from media, which would fail. Resolves: rhbz#738577
| | * | | Fix a typo (#794504).Chris Lumens2012-02-171-1/+1
| | | | |
| | * | | Add support for network --device=link in stage2 kickstart (#790332)Radek Vykydal2012-02-171-1/+9
| | | | | | | | | | | | | | | | | | | | Without it we hit KickstartValueError in stage 2.
| | * | | Don't set the pmbr bootable flag on Macs, whether booted via EFI or notMatthew Garrett2012-02-161-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | Macs object to seeing the active flag set in the pmbr, even if they're booting via BIOS. Avoid doing that.
| | * | | Don't set GPT HFS+ partitions as bootableMatthew Garrett2012-02-161-1/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Setting a GPT partition as bootable changes it to the EFI system partition type, which is defined as FAT. Doing this for HFS+ confuses the firmware. Skip setting it bootable in this scenario - the firmware is fine with that.
| | * | | Mark HFS+ as fsckableMatthew Garrett2012-02-161-0/+1
| | | | | | | | | | | | | | | | | | | | | | | | | HFS+ filesystems will be mounted read-only if they're not clean, so get fstab set up correctly.
| | * | | Set default lang and create default locale files earlyWill Woods2012-02-163-4/+12
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | If you import locale before the file(s) get created, you end up with a traceback from locale that says: Error: unsupported locale setting And this happens no matter what the locale is. We used to create the locale files for the default lang in loader, so now let's have anaconda declare a default locale and build its locale data before we import locale.
| | * | | Generate connection UUID in inital ifcfg files created by anaconda (#705328)Radek Vykydal2012-02-161-0/+8
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Prevents using the same UUIDs for multiple installations when NM creates UUID for already existing ifcfg file based on hashing of configuration file name. Resolves: rhbz#705328
| | * | | Take in change of a binary name (brcm_iscsiuio -> iscsiuio) (#731761)Radek Vykydal2012-02-161-5/+5
| | | | | | | | | | | | | | | | | | | | brcm_iscsuio is symlink to iscsiuio currently.
| | * | | Add 'traceback' boot option for python-meh and libreport testingVratislav Podzimek2012-02-161-0/+4
| | |/ /
| | * | fix setattr in set_cmdline_boolPetr Schindler2012-02-141-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | It was missing an argument. Signed-off-by: Brian C. Lane <bcl@redhat.com>
| | * | Add _mounttype to HFSPlusMatthew Garrett2012-02-141-0/+1
| | | | | | | | | | | | | | | | hfs+ is called hfsplus in-kernel, so it's needed.
| | * | Add support for UEFI Mac installsMatthew Garrett2012-02-143-4/+27
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Macs are slightly special in terms of their UEFI implementation. The easiest way to handle boot choice is via the firmware boot picker menu, which doesn't make use of EFI boot variables. In order for something to show up here, the filesystem needs to be HFS+ and have some magic done to it. Handle this case by detecting that we're on a Mac and using an HFS+ /boot/efi partition rather than a FAT one. That lets us then install the mactel-boot package which will deal with configuring the bootloader.
| | * | Add support for HFS+ partitionsMatthew Garrett2012-02-142-1/+7
| | | | | | | | | | | | | | | | | | | | We'll want to actually create HFS+ partitions, so add the missing bits for that.
| | * | Remove networking configuration steps from linuxrc.s390 (#783227)David Cantrell2012-02-141-2114/+29
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | dracut does this now by using the ip= and rd.znet= parameters when booting the system, thus rendering everything linuxrc.s390 does in this area pointless. Some support code is there to ensure that NetworkManager does not drive over the already-configured interface.
| | * | Clear partitions' metadata when 'clearpart --initlabel' used. (#783841)David Lehman2012-02-132-29/+8
| | | |
| | * | Fix support for detecting existing mirrored lvs. (#734128)David Lehman2012-02-131-21/+45
| | | | | | | | | | | | | | | | | | | | | | | | We switched from using the lvs command to the pvs command in our udev rules, so now we have to count the _mimage_XX volumes to determine the number of stripes in a mirrored lv.
| | * | Don't put partitions into device.mapPeter Jones2012-02-101-4/+5
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Rationalle: <phcoder> pjones, mjg59: anaconda shouldn't include partitions in device.map <pjones> phcoder: oh? any particular reason why not, or is it just not a thing that should happen? <phcoder> pjones: we don't use partitions in device.map. We use the corresponding ioctl's. <pjones> oh, I see. okay, sure. <phcoder> pjones: such a line creates a disk "hd0\,1" <pjones> where the \ means "don't parse the comma as a partition"? <phcoder> hamzy: I've spoken with IBM guys. They are looking into as to why their system claims to be 100x40. Do you know a reliable way to detect these systems which claim wrong resolution? If so we can include a workaround <phcoder> pjones: yes. We need it because of IEEE1275 <pjones> hah, of course. <pjones> /pci@1e,600000/ <phcoder> pjones: these entries claim that BIOS exposes separate partitions as fully-pledged disks. <pjones> right, which is clearly not the case.
| | * | fix potential EFIGRUB infinite loopBrian C. Lane2012-02-101-1/+1
| | | |
| | * | finish ROOT_PATH changes in bootloader (#789169)Brian C. Lane2012-02-101-7/+7
| | | |
| | * | Be more verbose about upgrade failures (#735060)Brian C. Lane2012-02-103-10/+39
| | | | | | | | | | | | | | | | | | | | | | | | When an installed system cannot be upgraded it may be helpful to know if it was the product, version or arch that didn't match. This propagates that info up from productUpgradable so that it can be show to the user.
| | * | log tracebacks from importing formatsBrian C. Lane2012-02-091-1/+3
| | | | | | | | | | | | | | | | | | | | Sometimes a format module may have an error in it during development. It would be helpful to see the traceback to speed up debugging.
| | * | Skip setting PMBR boot flag on EFI (#754850)Matthew Garrett2012-02-091-2/+6
| | | | | | | | | | | | | | | | | | | | | | | | | | | | Macs won't boot with the PMBR boot flag set, and EFI systems won't need it set so skip it when the system is an EFI system. Signed-off-by: Brian C. Lane <bcl@redhat.com>
| | * | New version.Brian C. Lane2012-02-093-3/+7
| | | |
| | * | Set ONBOOT=yes for FCoE devices (#755147)Radek Vykydal2012-02-092-0/+15
| | | | | | | | | | | | | | | | Resolves: rhbz#755147
| | * | Add details to transifex.doc for branchingBrian C. Lane2012-02-081-2/+10
| | |/
| | * New version.Brian C. Lane2012-02-083-42/+64
| | |
| | * anaconda_optparse.py: a new OptionParser that also reads boot argsWill Woods2012-02-082-6/+135
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This commit adds AnacondaOptionParser, which is an OptionParser subclass that parses kernel boot arguments in addition to commandline arguments. It looks for boot args that match the parser's long options. You can also provide extra boot args to look for, e.g.: op.add_option("--kickstart", "ks", dest="ksfile") This patch also switches anaconda over to the new parser, which means we no longer need loader to parse boot args for us. NOTE BEHAVIOR CHANGE: anaconda now expects (but does not yet require) all boot arguments to begin with "inst.", and will emit warning log messages if you use the old forms (repo=..., ks=..., etc.).
| | * Add flags.set_cmdline_bool and flags.read_cmdlineWill Woods2012-02-081-19/+13
| | | | | | | | | | | | | | | | | | | | | | | | set_cmdline_bool replaces decideCmdlineFlag - it uses BootArgs.getbool. read_cmdline sets a bunch of flags based on the contents of cmdline - this code is basically just split from where it used to live at the end of __init__.
| | * flags.py: add new BootArgs() object for dealing with boot argsWill Woods2012-02-081-26/+73
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | This replaces Flags.cmdline (and Flags.createCmdlineDict()) with a new BootArgs class. BootArgs' parsing is slightly cleaner and it reads from other data locations (/run/initramfs/etc/cmdline for example). It's a subclass of OrderedDict, so things specified later on the boot commandline can override things near the start.
| | * flags.py: rework/cleanup Flags objectWill Woods2012-02-081-54/+48
| | | | | | | | | | | | | | | | | | | | | - make Flags a subclass of object - Drop the silly games with __dict__['flags'] in favor of directly setting attributes - Preserve the exception-on-get/set-of-unknown-attribute behavior
| | * fix serial console option parsing (#767745)Will Woods2012-02-081-8/+49
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Anaconda was expecting serial console options with only the baud rate after the comma, for example: console=ttyS0,115200 But this is also valid (see Documentation/kernel-parameters.txt): console=ttyS0,115200n8 So this patch adds a function to parse the options correctly, and improves the GRUB serial_command() method to use the parsed options wherever they vary from the default values. Based on a patch by Jan Stancek <jstancek@redhat.com> - thanks!
| | * run convertfs on upgrade (#787893)Brian C. Lane2012-02-082-1/+27
| | | | | | | | | | | | | | | | | | When upgradig to F17 the dracut convertfs script needs to be run on the target filesystem so that it can convert /bin, /sbin, /lib and /lib64 to symlinks into /usr
| | * check if stdout and stderr are the same in execWithRedirect and open the ↵Masahiro Matsuya2012-02-081-1/+4
| | | | | | | | | | | | | | | | | | | | | file only once in such cases Resolves: rhbz#750710 (cherry-picked from 7cba80ba8d8d934c4596404686b46887c398d028)
| | * Disable ipv6 on target system when using noipv6 option (#735791)Radek Vykydal2012-02-081-0/+17
| | | | | | | | | | | | | | | | | | | | | | | | The option can be set in boot command line, in kickstart file using network command, and in loader UI. The option takes effect only if there is no device with configured ipv6. Resolves: rhbz#735791
| | * New version.Brian C. Lane2012-02-063-6/+20
| | |
| | * Set the boot flag on the GPT PMBR (#754850)Brian C. Lane2012-02-032-8/+7
| | | | | | | | | | | | | | | | | | Some BIOS systems (LENOVO) have problems booting from GPT without the boot flag of the protective MBR set. This makes sure that it is alway on when using GPT labeled disks.
| | * Add missing _boot_description values for dasd and zfcp (#739620)David Cantrell2012-02-021-0/+5
| | |
| | * Select the same device for ksdevice=link in loader and stage2 (#760250)Radek Vykydal2012-02-021-1/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | In loader strverscmp(3) is used so the patch will cover vast majority of cases (although e.g. eth9 would precede eth10 in loader, whereas eth10 would precede eth9 in stage 2 if there were more than 10 devcies having link). Resolves: rhbz#760250
| | * New version.Brian C. Lane2012-02-013-3/+7
| | |
| | * Add a separate function to get an LV's VG name.David Lehman2012-02-012-1/+4
| | | | | | | | | | | | | | | | | | | | | It's a different key depending on whether you are after the VG name for a PV or an LV, so make it clear by using a separate function. Related: rhbz#772878
| | * util-linux-ng is now util-linuxBrian C. Lane2012-01-311-1/+1
| | |
| | * New version.Brian C. Lane2012-01-313-3/+13
| | |
| | * liveinst: canonicalize live-baseloop symlinkBrian C. Lane2012-01-261-1/+1
| | |
generated by cgit (git 2.34.1) at 2025-10-03 09:10:35 +0000