| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
s390x DASDS don't have bootable partition flags.
|
|
|
|
|
| |
This identifies that a device is part of a multipath, and builds an
mpath device for partitioning.
|
|
|
|
|
| |
This allows for the backing DiskDevice._partedDisk to be created
just-in-time, instead of when the instance is created.
|
|
|
|
|
|
| |
This cleans up the language in questionInitializeDisk(), as well as
making the disk description a parameter, so we can provide a better
description than parted would when one is available.
|
|
|
|
|
|
|
| |
Right now we identify disks as partedDisk.model, which looks really bad
when using DM-based devices. Add a description to DiskDevice,
defaulting do partedDisk.model, and use it in the UI, so that subclasses
can easily replace this with more useful text.
|
| |
|
|
|
|
|
| |
The udev code can be used for more than just storage - it can be used for
networking as well. Moving these methods out makes that a little more clear.
|
| |
|
|
|
|
|
|
|
| |
Add a dracutSetupString method to devices.py classes, this can be used
to ask devices used for / to return a string to add to the kernelcmdline
so that dracut can find /. For most devices nothing is needed, but for
iscsi some cmdline arguments are needed.
|
|
|
|
|
| |
This is a preparation patch for adding support for writing
the necessary dracut cmndline options to grub.conf
|
|
|
|
|
| |
This is a preparation patch for adding support for writing
the necessary dracut cmndline options to grub.conf
|
|
|
|
|
| |
When checking if a logical partition dependsOn() another device because that
other device is an extended partition also make sure they are on the same disk.
|
|
|
|
|
|
| |
One iscsi target / node can have multiple LUN's currently we would then set
it to autostart if any of the LUN's was not used for /, the correct thing
todo is to only set it to autostart if none of the LUN's are used for /
|
|
|
|
|
|
| |
Add functions to go from an iScsiDiskDevice to an libiscsi node and the other
way around. This is a preparation patch for adding support for writing
the necessary dracut cmdline options to grub.conf
|
| |
|
|
|
|
|
|
| |
Unfortunately, these values are not necessarily enclosed in quotes but
could still include spaces. We don't want the values in our results hash
to be lists, so we need to skip splitting for some keys.
|
|
|
|
|
|
|
| |
Network devices do not necessarily have a N: and S: entry in the udev info,
so we don't want to ignore those right out. Instead, move the decision up
into the caller and just have udev_parse_block_entry read all the info and
shove it into a hash.
|
| |
|
|
|
|
|
|
| |
This can be helpful in determining the cause of failed upgrade from log.
I used it when working on #499321. Can be kind of corner-case, but can't
hurt.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
When we ignore partitions because we ignore the disk, or because
the disk is part of a bios raid set (the case I hit causing me to write
this patch), we need to make sure lvm also ignores the partitions
and does not happily try to use PV's it thinks are on there.
|
|
|
|
|
|
|
|
|
| |
We do not want to bring arrays up (and esp not use, so not scan them) in
degraded mode, so add --no-degraded to the mdadm invocation.
Note we do not add --no-degraded to the mdadm invocation for the last
disk of the set, because doing so assembles the array in an inactive
(and thus unusable) state.
|
|
|
|
|
|
| |
This function takes a device name (e.g., sda) and returns the link name
that it matches in /dev/disk/by-path. Used to help people with many
similar named devices (hundreds of DASDs, for example).
|
|
|
|
|
|
|
|
|
|
| |
When a partition table is unreadable, anaconda displays a
window explaining the situation and asks you if you would
like to format the disk for use. The existing message would
only give the device node name (e.g., /dev/sda47). This
patch adds the drive model name and capacity to the message
so confused users might know which disk anaconda is talking
about.
|
|
|
|
| |
Just auditing the world for clarity, don't mind me.
|
|
|
|
|
| |
We have lots of strings that we display to users which are unclear or
contain poor usage and grammar. That's bad.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Parsing of /proc/scsi/scsi updated to work with latest zFCP.
Matching SCSI ID correctly with all of zFCP device, WWPN, and LUN.
Synchronization with udev_settle.
Error case is handled gracefully and reports user readable error message.
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
Reverse most of commit 9e4d2e76713c9c71dcdaf22767211c13be9b3668, because it
currently is not working and MD containers can for 99.9 % be treated as
regular mdarrays. The next patch from this set instead makes the necessary
changes to use the regular mdarray handling code, reducing the amount
of code.
|
|
|
|
|
|
|
|
| |
While testing isw bios raid using mdraid, I hit a backtrace caused
by all udev_is-foo() checks failing. This patch logs an error and
ignores the device instead of backtracing. I believe this may work
around quite a few "'NoneType' object has no attribute Name" bugs we
have open at the moment.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This should fix up a majority, if not all, of the 'cannot commit to disk
after 5 attempts' errors. I was hitting this today. The cause I found
to be active logical volumes still around when the storage code wanted
to commit changes to the disk. When a logical volume is active,
libparted is getting EBUSY when it tries to commit changes to the disk
and tell the kernel to reread the partition table. If the LV is active,
we need to deactivate it and the volume group before we start committing
changes to disk.
Adding a status() property to LVMLogicalVolumeDevice that looks at the
lv_attr field for an 'a' fixes the problem for me on F-11. All of the
code to down LVM devices is there, we just weren't checking the LV
status correctly.
|
|
|
|
|
| |
Include the lv_attr field in the lvm.lvs() hash table returned. Needed
to tell if the logical volume is active or not.
|
|
|
|
|
| |
'sparc' and 's390' are what are returned by iutil.getArch; however, we don't
actually support 31/32-bit installs on those machines.
|
|
|
|
|
| |
Add FCoE disks to the devicetree with a type of FcoeDiskDevice instead
of DiskDevice
|
| |
|
|
|
|
|
|
| |
Add the ability to identify if a disk is an FCoE device and to
get the identifier of the disk and the name of the NIC used to
connect to an FCoE disk
|
|
|
|
|
| |
Write out an /etc/fcoe/cfg-eth# file for each nic used to connect to
an FCoE SAN during the install.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch adds support for using FCoE during the installation. This patch
merely lays the initial ground work, there is more work todo:
- The system will not boot without manual help after the install, as
dracut / mkinitrd do not support FCoE yet
- If FCoE is not used for / but for example for /srv, then information
about the nic used for FCoE needs to be written in a to be defined
config file in the system, and rc.sysinit needs to be thought to read
this file and bring up FCoE SAN's / Fabrics not used for /
- kickstart support for FCoE still needs to be done
|
|
|
|
|
|
|
| |
Fix of this UI flow: doing storage reset after root partitions discovery (e.g.
in partitioning step) and then going back and selecting upgrade. In place of
using root partition device objects obsoleted by storage reset, find root
partition devices of new device tree.
|
| |
|