Mount cache in image_create() image_restore() and cache_fetch().
Remove init_cache() and ensure /opt/opengnsys/images/ exists
within the cache mountpoint if it does not exists in cache_mount().
Add compute_md5 function in src/utils/fs.py
Remove identical md5 functions from src/live/ogOperations.py and
src/utils/tiptorrent.py
Move error checks from ogOperations.py into compute_md5 function in
src/utils/fs.py
Mount the system partition in readonly mode and check for the
hiberfil.sys file if the target system is a Windows.
Fail the image creation process if the target system is hibernated.
python-libfdisk does not close file descriptor to /dev/sda after completing
partitioning. This results EBUSY errors when formatting partitions with mkfs
in newer kernels. Encapsulate code to partition in method so python garbage
collection knows ctx objects can be release then close file descritor to
/dev/sda.
ogRest is not accessible from _partition(), remove check to ogRest.terminated,
actually no need to terminate inmediately when formatting is ongoing, better
leave things in consistent state when stop command is received.
Reuse os_probe() function from probe.py and change the fallback
case to "unknown" to prevent OS id mismatch in ogserver.
The OS id mismatch causes the images to stop being associated with
partitions after the /refresh command.
Add python implementation of the legacy ogWindowsRegisterPartition
function.
This function configures the system partition to be considered the
new C drive after a Windows image restore.
The drive letter configuration is stored in the SYSTEM hive of
the windows registry. The node MountedDevices contains all the
configuration as key-value pairs.
The C drive key \DosDevices\C: contains a different value based on
the partitioning type.
GPT
the value is DMIO:ID: followed by the 16 byte GUID of the
partition.
Example: DMIO:ID:\xc9\xfc\x1c\x86\x13\x11O\t\xa2\x9c{/\xf1\xdf\xe4)
MBR
The value is a little endian byte sequence with 2 parts.
The first 4 bytes correspond to the disk id. The second part is
the byte offset start of the partition in that disk.
Example: \xe1\\\x9cP\x00\x00\x10j\x18\x00\x00\x00
If we format the MBR value in a more readable way we get
e1 5c 9c 50 00 00 10 6a 18 00 00 00
In this case the disk ID is 509c5ce1.
The partition offset is 186a100000.
This patch adds the following helper functions to:
- get_disk_id_bytes(): to obtain the disk identifier.
- get_part_id_bytes(): to obtain a partition identifier as UUID
for MBR or DMIO:ID format for GPT.
- get_sector_size(): to query the sector size of a specific disk.
Read /sys/class/block/{device_name}/queue/hw_sector_size to obtain the value.
This is MBR specific.
- get_partition_start_offset(): to query the start sector of a specific
partition and disk. Use sfdisk with the -J argument to get fdisk data in
json format. This is MBR specific.
Add 'efi' key into the refresh payload. The value for that key
has the following structure:
'efi': {
'entries': [
{
"order": 0,
"name": "Boot0000",
"active": false,
"description": "grub"
},
{
"order": 1,
"name": "Boot0001",
"active": true,
"description": "UEFI: PXE IP4 Realtek PCIe GBE Family Controller"
}
]
}
If the client is not a EFI system it won't add the 'efi' field.
If an entry is not in the boot order it won't have the 'order' field.
fe40f9c5 ('src: add POST cache/fetch method') broke unicast cache restore.
(2024-09-09 10:05:22) ogClient: [ERROR] - name 'image_name' is not defined
Traceback (most recent call last):
File "/opt/opengnsys/ogClient/src/ogRest.py", line 175, in image_restore
payload = ogRest.operations.image_restore(request, ogRest)
File "/opt/opengnsys/ogClient/src/live/ogOperations.py", line 520, in image_restore
self._restore_image_unicast(repo, name, partdev, cache)
File "/opt/opengnsys/ogClient/src/live/ogOperations.py", line 242, in _restore_image_unicast
image_path = f'{OG_CACHE_IMAGE_PATH}{image_name}.img'
NameError: name 'image_name' is not defined
Implement configure_fstab() as a replacement of ogConfigureFstab.
Create src/utils/fstab.py to implement the main fstab
configuration functions. Define two fstab helper classes,
FstabBuilder and FstabEntry.
FstabEntry
Represents each line in the fstab file.
Has the values: device, mountpoint, fstype, options, dump_code
and pass_code.
FstabBuilder
Contains a list of FstabEntry. Handles loading of a
preexisting fstab file and the serialization of multiple
FstabEntry into a file.
The fstab configuration has 3 main steps:
Root partition:
- Update the device field with the device where
the new system is installed.
Swap partition:
- Preserve all the swapfile entries in every case.
- If the filesystem has a swap partition: update the device field
in the first fstab swap entry and remove the rest swap entries
pointing to a swap partition. Only one swap partition is supported.
Create a new fstab entry if no preexisting swap entry exists.
- If the system has no swap partition remove every swap partition
entry.
EFI partition:
- Update the device field of the EFI fstab entry if it exists.
Create a new fstab entry if no preexisting EFI entry exists.
Add get_filesystem_id to disk.py to obtain the UUID.
Define every device field as a UUID. That method is more robust
than a plain device path as it works after disks being added or
removed.
instead of
Checking partition "Microsoft Basic Data"...
Checking partition "EFI System"...
show
Searching EFI partition...
EFI partition found at /dev/sda1
and refer to EFI partition consistently in logs.
if image already exists in the cache, skip a second checksum validation.
log shows duplicated entries:
Verifying checksum for example.img, please wait...
Checksum is OK for example.img
Verifying checksum for example.img, please wait...
Checksum is OK for example.img
because tip_check_csum() is called twice in this case.
Restoring an updated image file (with different checksum) could fail while the
old checksum file could remain in place.
Remove image and checksum file before fetching the new files.
.full.sum file in the local cache contains the local checksum for this file,
this checksum is calculated by tip_write_csum(), therefore, there is no need to
calculate this checksum again from the image file in the cache, use this
checksum content instead.
if .full.sum is not available, then cache is inconsistent, tip_check_csum()
is called after checking if image file exists, raise an exception.
checksum could mismatch in two situations:
a) new image version (checksum is different)
b) image is corrupted
in both cases, a new fresh image needs to be retrieved.
this speeds up checksum validation.
Alejandro Sirgo Rica
OpenGnSys Support Team