| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The main problem here is that get_partition_by_name() has been
removed from partedUtils.py. The functionality is now provided
by the getPartitionByPath() method on a parted.Disk object. The
old function in partedUtils would iterate over the PedDisk
collection and each partition manually, whereas now we have to
call getPartitionByPath() for each parted.Disk we have.
request.drive has been used inconsistently throughout the existing
storage code and only started to show up when I was moving things
from anaconda to pyparted. What this patch does is ensure that
request.drive on RequestSpec objects is always a list or None.
There are a few other changes in the storage code to handle the
request.drive variable always being a list, but I think it should
be pretty straightforward.
|
| |
|
|
|
|
|
|
| |
Use the getFirstPartition() and nextPartition() method to
iterate over partitions so we can see the free space as
represented by libparted.
|
|
|
|
|
|
|
| |
Rather than calling disk.nextPartition() over and over, we can
get a hash table of all partitions on a disk via disk.partitions.
Iterate over the values() of this hash (the keys correspond to
the partition number).
|
|
|
|
|
| |
The property name for Device objects on a Disk or Geometry is
'device', not 'dev'.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
1) Iterate over partitions with a for loop over
disk.partitions.values() rather than calling next_partition()
2) Call getFlag() rather than get_flag()
3) Call setFlag() to enable a flag, call unsetFlag() to disable
a flag.
4) Reference the active property rather than calling the
is_active() method.
And other fixes.
|
|
|
|
|
| |
This function is now provided as a method on parted.Partition
objects.
|
|
|
|
| |
Now provided by parted.Disk.getLVMPartitions()
|
|
|
|
| |
Now provided by parted.Disk.getRaidPartitions()
|
|
|
|
| |
Now provided by parted.Disk.getLogicalPartitions()
|
|
|
|
| |
Now provided by parted.Partition.getDeviceNodeName()
|
|
|
|
|
| |
The parted.Partition object provides getSize() to return the size
of the partition in a human-readable format.
|
|
|
|
| |
This function is just a pass-through for isys.readFSType().
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch replaces the iscsi.py code with new code using libiscsi, getting
rid of the awefull configfile mangling to set auth info hack, and in general
making it much cleaner.
Fixes:
-Report error when we cannot add a disk instead of silently failing (461830)
-Report error when trying to use iscsi from kickstart without having an
iscsiname directive first (463156)
-Better feedback while scanning iscsi disks
-Make ibft work with chap and 2-way chap
-Make ibft configured disks available for upgrades and in rescue mode
-Make ibft boot work even when iscsi-initiator-utils is installed after the
kernel in the transaction
Note:
libiscsi is available in Fedora now
|
|
|
|
|
| |
data may not exist in the snapshots hash. If it isn't there, return
to caller.
|
|
|
|
|
| |
Also, make LUKSDevice's passphrase member into __passphrase so it
doesn't end up in anacdump.txt.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
anaconda kills and restarts iscsid each time an iscsi / zfcp disk gets added,
resulting in messages like these on tty4:
<28> Nov 6 11:30:06 iscsid: connection1:0 is operational now
<28> Nov 6 11:30:07 iscsid: iscsid shutting down.
<28> Nov 6 11:30:08 iscsid: iscsid with pid=1359 started
<28> Nov 6 11:30:09 iscsid: session
[iqn.2005-03.com.max:testdisk,192.168.1.100:3260] already running
<28> Nov 6 11:30:09 iscsid: connection1:0 is operational after recovery (1
attempts)
Esp. the "connection1:0 is operational after recovery (1 attempts)" clearly
indicates that this is not a good thing todo. iscsid is designed as a kernel
helper daemon, which when iscsi is used should be started and never stopped
again.
Note that this means iscsi.startup() also will not get called during the
partitionobjinit step, this is a no-op the first time as then the initiator
name has not been set yet, and if we go back and forth over this step we do not
want to start / stop iscsid, just as we don't want to do that for each iscsi
target added.
|
|
|
|
|
| |
kickstart and upgrading. This patch correct that behavior by calling it when
we search for devices.
|
|
|
|
|
|
| |
We can actually check earlier on the live image how much space is needed
and so we can pass that along. This reuses a string from later in the file
so that we don't break the string freeze
|
|
|
|
|
|
|
| |
There was a regression in the patch for bug 465240 (commit 9c55aab9)
that caused us to not only filter out locked PVs, but also newly
created unencrypted PVs whose device name matched that of a just-
deleted unlocked encrypted PV. Confused yet?
|
| |
|
|
|
|
|
|
|
|
| |
This arises from the fact that we determine whether or not a device
is a PV by looking at the partition type instead of probing the
device's contents like with other filesystems/ptypes. So, if we have
a PV that the user chose not to provide a passphrase for, we should
not include it in the list of available PVs.
|
|
|
|
|
| |
ext4 was detected only on lvm or raid or encrypted partitions.
Fw port of rhel5's #466868 and #465248.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
What is happening here is that with / on iscsi we do not want to try and start
the node(s) corresponding to the raid / logvol / partition used to autostart
because they are already started and the node database may even contain wrong
info causing problems as described in bug 437891. However for /somedir on iscsi
we do want to set the node to autostart, as there the node will not have been
started from the initrd, so without starting it we won't have a connection to
the disk for /somedir.
Thus this commit sets node.startup to automatic for non / containing iscsi
nodes only.
|
|
|
|
|
|
| |
We only prompt for one passphrase for new devices. It gets used for
all of them. If there are preexisting devices for which we have keys
we also offer the option to add the new passphrase to the old devices.
|
| |
|
|
|
|
| |
Do not, however, put the passphrase in there since it would be plaintext.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
since they might be started before there are any LVs. (this fixes dmraid.)
|
| |
|
| |
|
| |
|
|
|
|
| |
* po/sk.po: Typo fix.
|
|
|
|
| |
check that it's not our install image.
|
| |
|
| |
|
| |
|
|
|
|
| |
Use of the /dev/mapper/VolGroupXX-LogVolXX node allows us to rely on the basename being unique.
|
|
|
|
| |
Scan for encrypted devices while reading the initial partition layout from disk. Prompt for passphrases and include encryption information in the RequestSpec.
|
|
|
|
|
| |
This fixes mounting disks by UUID in upgrades and probably rescue mode
as well.
|
| |
|
| |
|
| |
|
|
|
|
| |
When scanning partitions for LUKS headers, we first check if libblkid says the partition has a RAID signature (we cannot rely on partition type). If it does, we assume that any LUKS header present is spurious.
|
| |
|
|
|
|
| |
The user is prompted to enter a passphrase for each device as it is encountered.
|