Page 1 of 38

HDDSuperClone

Posted: Wed Mar 02, 2016 7:40 pm
by maximus
It's time to officially name it. The new Linux cloning tool will be named HDDSuperClone. It is currently in alpha testing, and for the most part functionally stable. There is much work to be done before a beta release of the free version, but I just wanted to make sure everyone knows I am still working on it :)

http://www.sdcomputingservice.com/hddsuperclone

HDDSuperClone

Posted: Wed Mar 02, 2016 7:58 pm
by maximus
I believe I have a good skipping algorithm for it. It learns as it goes and does the best it can to adjust to what it thinks may be weak/bad head so that it can get the most good data first. But I have also been playing around with trimming vs splitting (different than ddrescue splitting). Trimming means taking a larger bad block and reading it from one end to until a read error, and then reading it from the other end until a read error. This is the same as ddresuce trimming. My version of splitting involves taking a larger bad block and dividing it up into smaller chunks, the first pass is divided by 4 and second pass is divided by 16. Maybe I will rename it dividing instead of splitting. Anyway, it depends on the situation as to what is more efficient. For the basic free version with SCSI passthrough and not using any timers, trimming is most efficient, just as it is in ddrescue. But the use of timers and ATA passthrough can make dividing more efficient. It's fun trying to figure all this stuff out :)

HDDSuperClone

Posted: Thu Mar 03, 2016 3:47 am
by pclab
Excellent work.
Looking forward to test it.

HDDSuperClone

Posted: Thu Mar 03, 2016 7:12 am
by abedalkareem
Good Job

HDDSuperClone

Posted: Thu Mar 03, 2016 11:17 am
by LarrySabo
I salute your dedication, perseverance, determination in this project. I hope your considerable investment in time and effort is appropriately rewarded.

HDDSuperClone

Posted: Thu Mar 03, 2016 8:53 pm
by maximus
So I just upgraded my main drive from a SanDisk 256GB SSD to an Samsung 500GB SSD. I first tried to clone using Clonezilla (thought it would be faster), but it failed on one of the partitions. So I got crazy and used my own HDDSuperClone tool to clone it, which took just over 31 minutes. As I am typing this running on the new drive, I can say that it seems at least the basic version has passed a large scale data integrity test.

HDDSuperClone

Posted: Fri Mar 04, 2016 5:14 am
by pclab
Testing on yourself already??? eheheheheh
That's a major test.
Mad Scientists usually try their cure to the world disease on himselfs first ahahahhaha

HDDSuperClone

Posted: Fri Mar 04, 2016 9:56 pm
by Jared
Awesome work. I'll have to set up a test rig using some cases I've already finished and make some comparisons. Maybe even a YouTube video or two.

HDDSuperClone

Posted: Sat Mar 05, 2016 8:24 am
by maximus
Jared wrote:Source of the post Awesome work. I'll have to set up a test rig using some cases I've already finished and make some comparisons. Maybe even a YouTube video or two.

The first release will be the free version that only uses SCSI passthrough (would also probably unlock ATA passthrough for testing purposes). And the first thing I will need tested is the ability for it to detect a device fault and exit. So if you have cases where the drive needs to be power cycled that would be ideal. For this it would help for you to keep track of where some of those bad sectors were so you could jump right that area instead of waiting for it to get there. These drives must also be able to be seen by Linux and not fault out as soon as Linux touches it. My problem is the only drive I have that will device fault does so as soon as Linux touches it, so I wrote the code on a bit of faith (plus a few reports from people using my ddrescue passthrough patch).

After that I am curious as to how well the head skipping algorithm works on some real world cases. I only have one that I used to test it. Just remember that the passthrough is still slow at processing bad sectors, unless the drive supports error recovery control and you change the read timer :)

HDDSuperClone

Posted: Sat Mar 05, 2016 1:30 pm
by maximus
Looks like I am done playing for today. Just noticed that my 2TB Hitachi in my test machine was reading very slow. Checked the smart data and plenty of reallocated sectors. The strange thing is that neither the computer BIOS or Linux gives any messages about the drive failing, even though it should not pass a smart test. While it does not contain super critical data, I would rather not loose what is on it. I guess it is somewhat good timing, as I just pulled an identical Hitachi out of my main computer and replaced it with a WD 4TB, which means I have something to clone it to. So it looks like I am testing on myself again, this time for a real failing drive :shock: