[mythtv-users] python filewriting and partition boundaries

[John Pilkington]

On 03/04/13 06:28, HP-mini wrote:
> On Tue, 2013-04-02 at 21:22 -0400, Raymond Wagner wrote:
>> On 3/31/2013 12:57, John Pilkington wrote:
>>> It seems to me that the most unusual thing about what I have done with
>>> this box is using a python script to cut and concatenate video files.
>>> The script is at MythDVBcut on the wiki.  In the scripts as posted it is
>>> used only for HD content.  It doesn't include any checking for partition
>>> boundaries, because I assumed that this would be done automatically.
>>
>> When you mount a filesystem, the filesystem translates your file
>> accesses to its own internal structure on the physical media.  You do
>> not directly access the physical media, so there is no possible way for
>> you to corrupt the physical media, or the partition data stored therein.
>>    The only way you could have corrupted it yourself would be if you
>> logged in as root, and wrote directly to the device node.
>> _______________________________________________
>
> You mention the first corrupt file system was NTFS.
> Until recently.. writing to an NTFS file system from linux was not
> recommended.
>
> Possibly the NTFS partition/HDD was removable &/or shared with windows
> box.
> Could you blame the drive or adapter or windows?
>

That was several years ago, but it made me pause before trying anything 
similar again.  It was a dual-boot box with two discs and the NTFS space 
looked inviting, but a few days after I started using it - as target for 
demux operations, IIRC - everything stopped working and boot attempts 
just hung until I unplugged that drive.  It's still in the case, unused. 
  Perhaps some of the data on it was important, but I've lived without 
it since.

Of course it might be another bad diagnosis.  Even without that disk in 
circuit the first boot attempt usually fails and asks for a system disk. 
  Second attempt works.  Centos 5, Myth 0.24 something, Pentium 4 space 
heater.

And thanks, Raymond, for your reassurance.  I felt some doubt about 
multi-GB file operations on a 32-bit machine with no prior indication of 
file length, but the root cause of the recent problem seems to have been 
the unintended removal of the LVM management tools.