HDDSuperClone

[nissimezra]

It's not too late to consider going with a hardware/software locked solution. A single-board computer with the software built in and custom OS wouldn't have any compatibility issues because it'd be running exactly what you put on the machine.

Now I know who you are. You are the nice guy that answered my questions on youtube regarding PC3000 Course cost 5500$. Thx for the advice!

Now to the topic.

I agree with half of what you are saying.

@maximus you are wasting time on this, who cares if it's Lubuntu 14.4 or 17 or whatever OS there is? No one. You are selling a data recovery tool not an operating system. No one is going to use it as an operating system anyway! It can be used as a bootable CD / ISO and thats should be enough.
Or just an installer with the operating system, can be used as second drive or dual boot. remember most of the people buying it should have some knowledge in computers, you are not targeting people with basic knowledge.

@Jared you are making thing unnecessary expensive for him. The cost of the Machine is at least 200$ for him, thats if you are talking about a simple computer with memory cpu etc. His goal is to sell it around 200$.

@maximus my opinion sell it as is, 14.04 is fine.

Good luck

 

[maximus]

[post]10546[/post] It's not too late to consider going with a hardware/software locked solution. A single-board computer with the software built in and custom OS wouldn't have any compatibility issues because it'd be running exactly what you put on the machine.

Going the hardware route would just further complicate things, cost more, take up more of my time, and add more limitations. So I think I have ruled that out.

After some more attempts to build on 16.04.1 (I can’t even get 16.04 to boot on my one computer), I have decided that at this time I am sticking with 14.04 for my building and the live cd, as there have been no complaints so far. I am also not officially supporting any Ubuntu above 16.04.01 or any Linux build with a kernel above 4.4. With kernels above 4.4 the ata passthrough does not return ata registers, but it will still work for cloning as it uses the scsi sense data instead for error detection. I still need to figure out what I am doing wrong (or not doing right) that causes it to lock up with kernel 4.13 on one of my computers when performing a simple identify command. Hdparm can do it without locking up, so all hope is not lost there, plus it is only that computer that locks up, another one is fine. I am not saying that it won’t work on kernels higher than 4.4, but at this time I am not going to support it if something isn’t working.

As far as I know, the direct modes should still work on the higher kernels without issue since I am bypassing the kernel drivers altogether, although I have not really tested much. I will find out more on that when I produce the pro version.

Now it’s time to work on the driver issue I have with kernels 4.11 and newer, but that one seems solvable with a conditional code change, just had to spend a day or so trying to find the information about what changed in the kernel.

 

[maximus]

[post]10558[/post]
Real world job, I received this hard disk from a computer store.
SMART report failure predicted

ddrescue GUI did not load it, it freezes

https://imgur.com/a/rcUN6GR

So far so good but it keeps slowing down every day.
Now it shows 99.61 percent copied, is this result accurate? if it is 99.6% it's more than enough to extract the data.

The first 97% copied within 30 hours (about) it's very good. Can this result be trusted?
The remaining time is not even near accurate.

I will keep you posted about this job.

It looks like it just got into the trimming phase. The drive definitely has a bad spot, and now that it was is being processed. That is why it has slowed down, the algorithm did its job and got the data from the good areas first, and now it is grinding on the bad spot. Depending on how bad it is, it could take a very long time to process the bad spot. The speed can change depending on the exact nature of the spot as it goes along. The remaining time is calculated from the read speed of about the last 5 minutes or so of run time. Assuming you keep the progress log, you can always stop it to see if you can get enough of a recovery, and if not you can resume and see how it goes. The drive could die while grinding on the bad spot.

 

[maximus]

[post]10558[/post] Now it shows 99.61 percent copied, is this result accurate? if it is 99.6% it's more than enough to extract the data.

I would just like to point out that the bad spot I mentioned in my previous post looks like it may span across about 10% of the disk (75GB). That is assuming the drive has 4 heads, and doing some math on the skip runs and run size. That is just a guesstimate by what I can see from just the screenshot. That could potentially affect many files, depending on the nature and size of the files.

 

[maximus]

I know nissimezra made a post after mine, I saw it once a few days ago. So what happened to it? I think I couldn't see it the first time unless I logged in, but now I don't see it at all.

 

[LarrySabo]

Jared mentioned losing a day's worth of posts a few days ago when he had to restore from backup after a WP plugin update. Perhaps that was among the collateral damage.

 

[Jared]

Yeah, sorry about that. Had to do a rollback to 24 hours before after installing a plugin totally botched up my site. It was a pretty quiet day, but there were around a half dozen posts that got lost.

 

[maximus]

So right now I am playing around with a couple Hitachi 300GB SAS drives that have 520 byte sectors. They read fast with my card, about 120MB/s. But writes are slow at my default cluster size of 128, the speed is only 14MB/s. I can safely get up to a cluster size of about 504 (something about how linux is handling the drive is limiting it more than it should) with a read speed of about 35MB/s. But no matter what the drives click about every 5 seconds when writing. It is not a consistent time, sometimes a second quicker or slower, but both drives do it. And I have heard both drives do some crazy head movement for a few seconds on occasion. Is this normal for SAS drive writing? I am just wondering if there is either something not good with the drives (they are used), or something weird with either my card or the way lilnux is writing to them.

 

[nissimezra]

I know nissimezra made a post after mine, I saw it once a few days ago. So what happened to it? I think I couldn't see it the first time unless I logged in, but now I don't see it at all.

Hi Maximus,

Yes one of my post is missing but I don't remember what I wrote it was probably an answer to your post.
No worry I will finalize what I started with now

3 days after the post I aborted the scan at around 99.75% and the result were surprising.

I took the copied drive and loaded into win 10, win 10 ask to repair it, after that I had all partitions working, (remember the drive showed unallocated) Users folder were good I copied all the files easy. Note some images were corrupted like when scanning drive without partitions.

One thing to notice; after that I tried to continue the scan using the log file but it didn't work, I received a message that the log file need to be repaired first.

After that I manged to load the drive on ddrescue guy, after 10 days it finished 75% only. We can assume that the drive health decreased.

Thats all for this one, I'll keep you update if I'll have more jobs or use

 

[maximus]

In regards to the Hitachi SAS drives I was working with, I still don’t know how much of the weirdness had anything to do with the OS or card, but I did find that formatting one to 512 bytes per sector and letting the OS do the writes without scsi passthrough (needed for writing the odd sector size), the drive seemed to operate more normally and at a good write speed. So while it is possible to write to the 520 byte drives, it can be noticeably slower. On a side note, those drives get damn hot! I put a thermocouple on the outside of one and it was 132F! No wonder it hurt to grab it!