On 04/21/2017 08:47 PM, Phil Susi wrote:
On 4/20/2017 10:07 PM, Wang Dong wrote:
But the flag cannot be read with the presence of the filesystem previously.
That's why the change.
How so? The flag is in the partition table. What data is written to
the partition has nothing to do with the flag. Whether you have a
filesystem or not, the flag is still there in the partition table.
▷ ▷ if ((strncmp(PART_TYPE_RAID, str, 6) == 0) &&¶
451▷ ▷ (ped_file_system_probe(&part->geom) == NULL))¶
452▷ ▷ ▷ ped_partition_set_flag(part,
PED_PARTITION_RAID, 1);¶
453▷ ▷ else¶
454▷ ▷ ▷ ped_partition_set_flag(part,
PED_PARTITION_RAID, 0);¶
455¶
456▷ ▷ if ((strncmp(PART_TYPE_LVM, str, 6) == 0) &&¶
457▷ ▷ (ped_file_system_probe(&part->geom) == NULL))¶
458▷ ▷ ▷ ped_partition_set_flag(part,
PED_PARTITION_LVM, 1);¶
459▷ ▷ else¶
460▷ ▷ ▷ ped_partition_set_flag(part,
PED_PARTITION_LVM, 0);¶
Maybe I did not put it clearly. The flag is on the partition table. But
it cannot be read with a filesystem existence.
The code above is from libparted/labels/dasd.c, function dasd_read(),
master branch.
As I also tested it, with a filesystem, when I set the flag in
interactive mode, I can read the flag with "print".
But I can not get it after I re-read the device. And I can verify the
flag's existence with other tool.
If I set the flag in script mode, I cannot read the flag either. That
means user cannot see the flag, although it exists.
If I missed something, please let me know. Thanks in advance.
Nothing goofy. Just the property of the lvm/raid. Because when to
make a partition to lvm/raid, the filesystem will be ripped out.
Not by parted it won't. We just update the partition table, and it is
up to the user to mess with the filesystem. If they don't run pvcreate
on the partition after setting the lvm flag, then the filesystem is
still there and works just fine. These flags simply don't matter in
practice, and so I never bother setting them even though I do use raid
and lvm.
--
Best regards. Wang Dong
