Unfortunately, I probably lost the original Feb14-image.log. I think it disappeared when I tried repairing it from hddsc though I can't remember exactly (or has changed into a different file name).
Yesterday I continued Phase 2 with the "Feb14-image.log.Phase 1.2021-02-14_11.13.08.617101.bak" log file.
I think it still works fine (or not?) despite it only recorded Phase 1 logs even though the actual process was already on Phase 2 on the time the power outage shut everything out.

I also noticed that "Feb14-image.log.Phase 1.2021-02-14_11.13.08.617101.bak" (the one im currently using) doesn't automatically create another .bak file, in case another interruption shuts my system again.


You mean factory default skip size? the one that's 4096?

I'd viewed the logs as pretty much all the quick data has been read on pass 1- save fringe data, the rest is still good and can be read but it's slow. A small increase in wait time may reduce/remove phase 3. I'd prefer 1 pass backwards to get the lot rather than 1 back picking up very little and phase 3 forward grabbing the slow stuff, small drive only 160GB left factored in too if it was a 1TB drive with only 200GB on it then risking reading backwards would be stupid.

_________________
Properly, Quickly, Cheaply - Pick any two.

Why does everyone always want to skip phase 2? If the algorithm works as expected, phase 2 usually doesn't even take that long, and it gets the trailing edge of the over-skipped data. Yes, it is probably only about 1.64% according to the last known log (assuming 50% per head), but it shouldn't take that long to get that last bit of known good data.

After phase 2, well, it is trial and error after that in this case. I wouldn't even rule out disabling read ahead. I am not a big fan of reverse, but it couldn't hurt to try. But at least if phase 2 is finished you know you got (almost) every bit possible out of the good head.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

I mean the 16384 you set originally. Actually once you are done with phase 2 and try phase 3, you can play with the settings more, so you can set it to whatever makes sense. In phase 3 the skip size is absolute, and does not change on its own, you set it to what you think works.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

Confused : It's phase 3 Id hope not to need with the rest of the data being read in phase 2.


I suspect given a long enough time out this could well result in a 99%+ recovery, I don't think that heads dead yet. Interesting one, so much easier if you can hear and feel the drive.

_________________
Properly, Quickly, Cheaply - Pick any two.

Phase 2 is a compliment to phase 1 with the self learning head skipping algorithm. Whiles phase 1 seems to do a good job, it will have some over skipping, and phase 2 is designed to get the over skipped data. After phase 2, the remaining data should all be in the bad head or bad spot. Then it is time to play with settings if needed, but if playing with the settings, phase 1-2 should be disabled because you do not need them again. Phase 3 (rate based skipping) and/or phase 4 (no skipping) should be used until there is no more non-tried data, or until giving up. If you want a backwards pass, then enable reverse with phase 3 or 4. The current position may need to be changed as needed.

To sum it up, phase1 and phase 2 are for skipping the bad data (bad spot or bad head). Messing around with settings in the middle of phase 1 or 2 can screw up the skipping algorithm. After that, everything else is meant to dig into the bad, and messing with the settings may be needed to get the best results.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

I would like to add that after the initial phase 1 and 2 are completed, it is possible to utilize those phases again with modified settings, to dig into the bad. But you would need to adjust both the skip size and the max skip size to the desired values, otherwise the program could stop due to a skip reset.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

I'll let the OP get the data off.

If the OP is up for it afterwards and the drive hasn't died we could run a couple of tests. I'd like to see what happens with a single phase 1 pass with the timeout set 3-5K.

_________________
Properly, Quickly, Cheaply - Pick any two.

All I want to see is normal completion of phase 2. Then it is game on. I do think that if the drive survives long enough, this could be a 99.9% recovery.

Are you looking at the timing bytes in the log? Past 7GB (the first big bad head section) most of the reads in the bad head are in the 1 to 2 second range, a few under a second, almost nothing hitting 3 seconds. A single phase 1 pass with the skip threshold at 3+ seconds would likely skip some past the first 7GB, but then probably not skip much at all. Also, if there is skipping, the max skip would need to be adjusted to no more than 16 times the skip size, to prevent a skip reset (and program stop) due to the head skipping algorithm.

Okay, as I look closer, I can see some areas were there appear to be large chunks reading under 1 second, but probably slow. That is probably why it seems so slow, because it is not inducing skipping. But that is a good sign in a way. I think I would still let phase 2 finish normally. If anything I would drop the skip threshold to 500ms to induce skipping, to help phase 2 finish in a timely manner.

With the reads fairly consistent in the 1-2 second range, I am more curious if increasing the cluster size directly increases the read time or not. If it does, then nothing gained. But if the read times stay the same with a larger read size, then there is speed to be gained. No way to know without trying. This setting test should be done after the initial phase 2 completes.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

@Lardman

You mean starting off from Phase 1 but the reverse command is checked?

@Maximus
Hi, where did you find that 7GB from my progress log?


Where do I see the Read time display?

Also..... is it fine if I change the file extension "Feb14-image.log.Phase 1.2021-02-14_11.13.08.617101.bak" to ".log"
since I think it doesn't create any backup log file when I continued the clone with the ".bak" file.
Just making sure I can have something to hold on to in case an emergency happens again

From examining the log and using a calculator. It is advanced, not something I can explain.


It is under Current/Recent/Longest. That is the read time in milliseconds. When changing settings, the one to watch in real time is the first number (Current). It is the same with the rate, you watch the Current rate. You pay attention to what the numbers are doing before and after a setting change to see what difference it made. FYI the recent numbers are an average of the last 256 display updates, usually about 5 minutes.


Yes, but just to be safe, I would make a copy of it, and then rename the copy to something like "Feb15-image.log", and then use that.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

I started to continue cloning today, it's currently running at 75-100KB/s and seems to not reaching 500+ KB/s anymore.
Will update in the next few hours

settings: only Phase1&2 are ticked, skip size =16384, cluster size =default, skip threshold = default

This is taking so long, the remaining time is about 31 days
Did this all slow down because of the nasty power outage?

I'll try doing this on ddrescue if necessary but I'm still not sure which parameters to include. aside from " -n " or maybe "-E"

Update: the number of non-trimmed suddenly increased 2x from 896 to 1792.
It was 896 all the time until now.

I also think it's not rescuing much data, just sweeping quickly from 320GB position to 0GB @ Phase 2

That is not normal. I think something is screwed up if that is what you are seeing.

Okay, at this point, I am going to break down and agree with Lardman. Forget about finishing phase 2 with the current settings. Make a good backup copy of the current log. then make a backup copy of the backup copy (can never be too safe). Don't mess with the backup copy, as it's primary purpose is for reference if needed. Post the current log. Then reset the log status. Then it is time to start playing with settings.

From this point on, there is no "right" way to do things (but there are wrong ways). I think you should try a phase 3 run (only check phase 3 and nothing else). And try changing the cluster size. I don't have time right now to make any other suggestions. Without having the drive in front of me, it is hard to give instructions.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

I meant, it's not an abrupt process but the "Current Position" value was going from 320GB to a lower value gradually (right now it is 21.81GB) I think I worded it out wrong sorry.



It's alright, I really appreciate your time with this thank you very much nonetheless

log file



Should I still leave the skip size to 16384 ?

In that case it was working like expected, and should have finished in a reasonable amount of time. I think you caused it to take longer by changing the skip size back when you restarted the recovery... you were not supposed to do that until phase 3. Changing it in phase 2 messed with the skipping algorithm, and it had to relearn, which takes time and causes more reads in the bad head.

If you are going to abort the current phase 2 run, then yes, you will need to change the skip size, or it will end up being huge. 16384 is 8MB (16384*512). I am honestly not sure what the best skip size would be moving forward, that is were I would watch what was happening and make a judgment call.

FYI You might find that the data in the bad head is just going to read slow, and there is nothing you can do about it.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

I think I remember setting the skip size everytime to 16384 but there was this time that it turned 168000 without me knowing and I accidentally started it, but no worries.


alright, I'll continue the cloning process again with the same settings.
Settings: only Phase1&2 are ticked, skip size =16384, cluster size =default, skip threshold = 500ms

These slow reads that the source drive is experiencing reminds me the first minutes of the first clone attempt.
*hoping to recover to a speed of 50Mb/s again*

If you continue the phase 2, DON'T CHANGE THE SKIP SIZE! It is large for a reason! You will slow it down again and make it take longer. The log you posted has it at 499529, and it was working nicely.

_________________
http://www.sdcomputingservice.com
Home of HDDSuperClone and HDDSuperTool

Jeez, glad that I read this before I start.
But isn't this the same case as what you have mentioned here days ago?
Im slightly confused