So I've been doing major maintenance and one of those is a lot of moving and cloning
The history is somewhat relevant, on my now old SSD, a Win7 installation of 5 years began randomly losing it's MBR partition table integrity leading to the OS becoming unbootable and the disk recognized as unallocated, this was fixable with a linux tool TestDisk, sometimes the MBR boot sector is also damaged and it is recovered from the backup boot sector which TestDisk also does optionally.
The standard error that TestDisk would throw out is "Partition sector does not have endmark 0xAA55"
The partition table corruption would happen randomly, but sometimes it happened during video driver reinstall under Win7, one time it happened with a normal powerdown during the night.
Simultaneously I also investigated my power issues, It turned out my PSU was too weak for the amount of disks I wanted to run leading to weird behavior, but the stupid PSU never turned off when it was overloaded.
So I thought either the power or the old SSD is broken in HW, so I replaced the SSD. However this time I bought Samsung 860 EVO, which is a bit better than 850 (DDR4, newer controller, etc), because a few months ago back I got 850 EVO which I put a fresh installation Windows 10 on, but I wanted the old Win7 to be on the older 850 EVO, so I went into a big chain of cloning and moving. First time I did something like this in this scale, but I have some good experience in general how tackle new things and get the right approach, keep things logged in mind and not rush, but my troubleshooting skills with PCs in general come for help as well, so it not hard at all, it's just takes time and juggling things back and forth.
I probably rebooted the PC like 30 times and that's not the final number, heh. I started with making some image backups, then I directly used clone-disk-to-disk to clone Win10 from 850 EVO to 860 EVO. I don't think I disabled ntfsfix, I don't know about -e1 auto (adjust ntfs geometry) but I did disable resizing and I used "USE PARTITION TABLE FROM SOURCE". It was a first time success, I did some offline full chkdsk tests using the Win10 from 850EVO (had to temporairly change disk MBR ID so it would get online, then change it back, diskpart: uniqueid disk ID=8numHEX)
Win10 Seems to be running just fine, with no issues with any programs whatsoever so far, did 3 cold boots, and that's where I'm typing this message from right now.
It's a different situation with win7 completely ... note that I have since got a new PSU, and I didn't even plug in all the HDDs while doing this.
Motherboard: ASUS P9X79 (from 2012 I think, bought in 2013, it has UEFI, I run CSM and keep avoiding using most UEFI stuff to be on the safe side, so it's all MBR except data disks)
Clonezilla Version: stable - 2.5.5-38
OLD WIN7 SSD: Samsung 840 PRO 128GB
NEW WIN7 SSD: Samsung 850 EVO 250GB
WIN10 SSD: Samsung 860 EVO 250GB
The first time I cloned, with -k1 the cloned thing had the same problem after a second reboot, 0xAA55, in the BIOS this is reported as "NO OS WAS FOUND ..." had to go to TestDisk to fix, also bootsector recovered from backup boot sector, still wouldn't work, it kept saying DISK READ ERROR which never happened before on the old SSD 840 after fixing, so something was worse with the cloned situation.
There's many clones thereafter, I'll avoid explaining each one separately.
1. -k1 proportional partition - created a lot of chkdsk errors, programs broken, DLLs
2. -source disk/image partition - clone and image both created less chkdsk errors but programs were still broken, one of the times it was completely unbootable, the win7 autoloaded into recovery mode and reported system32\kdcom.dll was corrupted, same thing continue to happen after chkdsk fixed a bunch of files but not kdcom, so this was a failure again (used without -r but with ntfsfix and -e1 auto)
NOTE: THE FIRST BOOT IS ALWAYS SUCCESSFUL - Win7 STARTS AUTO DRIVER INSTALL FOR THE NEW SSD AND PROMPTS REBOOT NEEDED - AFTER THE SECOND REBOOT ISSUES START TO BE MORE OBVIOUS WITH ERRORS IN EVENT VIEWER, SCHEDULED PROGRAMS NOT STARTING, SHORTCUTS MISSING.
NOTE2: I USED SHA-1 SUM CHECKING AND MD5 FILE CHECKING IN ALL CASES AND ALL CLONES WERE SUCCESSFUL FROM CLONEZILLA SIDE
So it's all unusable, there's some guides on the net I see they suggest creating partitions manually before, so I'll try that, but the same guide also mentions the driver install for the disk so that appears to be normal.
Next I will try to create partitions manually ahead (using Win10) matching the original size and only clone data, I will also disable -r and ntfsfix and -e1 auto (is this ntfsfix?) - then I will manually resize the partition once it is cloned and checked if it works fine, that's one of the ideas right now.
I only noticed ntfsfix while looking at the fast scrolling verbose output in clonezilla otherwise I wouldn't recalled this thing is ran by default - I haven't figured out how to copy the full verbose clonezilla log from the RAMDISK to the USB, this has been the only big disappointment so far that this is not easier and basically found zero info on the web how to actually view it, but my debian and linux console skills are practically zero also.
After that, if it won't work, I'll go back to 840 SSD and disable all driver installation and see maybe if the driver install has some side effect that produces the NTFS corruption.
Even if the correct way is just around the corner for this Win7 Clone, it's not a problem, I do like to write such procedures in threads not only because it may help others but I find myself months or years later needing the same thing, it's also another backup to my notes which can get lost in the sea of text files.
Last edit: Anonymous 2018-04-19
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2018-04-19
Well, it's tough, I did the manual thing, recovered from image, it was successful in clonezilla, but didn't boot some 0x000007c error code, failure again, getting desperate ...
For some reason I didn't try ntfsclone or the other options, only partclone.
I just skipped to linux brainstorming, did a dd dev zero on the target drive, and just dd the 128GB onto the 250GB disk. First boot perfect, no ntfs errors, all programs started ok.
Second reboot, corruption prevented some programs from fully loading assets, photoshop that just worked, displayed an error about an unrecognized file, ntfs corruption events, please run chkdsk, and after that it's the same train with chkdsk fixing some 1000 files during boot, still the whole thing remains practically unusable as leftover corruption persists even if there's no corruption events.
I'm beginning to think it's some kind of software thing, either it's the side effect of the driver install for the new SSD or that the whole thing is broken because it's an old install and some drivers or programs could be messing something. But I have one of the toughest maintenances, there's no bloatware, this is not an average hollywood-style teenager type of installation with things thrown all over the place. But I do tweak a lot as well, all of my OSes are nowhere near default settings.
This more looks like it's going to have to be a fresh reinstall on the new SSD.
If I could only figure a way to do some logging to see what the heck is the driver installation doing, I can turn it off probably easily, on Win10 I've done tweaking as well, even heavier in some cases, but it's a much fresher installation but also more careful tweaking that is not too speculative, and it looks like I have done some thing to the auto-driver installation that I keep getting event errors about not being able to install driver for each of the partitions, I didn't even know I kept getting those errors for the 2 months it's been installe, maybe that made Win10 work so this thing should get investigated, I didn't really purposelly disable auto-driver install, I think it was just part of all the other tweaking and I forgot what was the reasoning, probably because it had to do something with windows update.
Last edit: Anonymous 2018-04-19
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Anonymous
-
2018-04-22
Well, I decided to do a clean re-install on the 250GB one, considering I just did Win10, I guess I have more fresh memory with windows tweaking and things may go faster than usual, probably justified at this point with a really old setup. Usually I'm determined to get to the bottom, but in my case it's not so critical, I can still keep using the old Win7 install alongside anyway until the new install is finely tuned up.
I could try preventing the driver install, but it's frustrating at this point, i've done full cleans on the SSD a number of times and don't feel doing it anymore.
Even if would be successful, it would be mixed feeling, I would do many reinstalls to move some programs onto the SSD fully since it's now all split to a HDD and messy with hardlinks.
If someone hits a similar scenario, the first thing to do would be to disable the auto driver install prior to cloning since after that things go south, or it has to do something with these specific SSDs aren't the similar enough under the hood.
Last edit: Anonymous 2018-04-22
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello
So I've been doing major maintenance and one of those is a lot of moving and cloning
The history is somewhat relevant, on my now old SSD, a Win7 installation of 5 years began randomly losing it's MBR partition table integrity leading to the OS becoming unbootable and the disk recognized as unallocated, this was fixable with a linux tool TestDisk, sometimes the MBR boot sector is also damaged and it is recovered from the backup boot sector which TestDisk also does optionally.
The standard error that TestDisk would throw out is "Partition sector does not have endmark 0xAA55"
The partition table corruption would happen randomly, but sometimes it happened during video driver reinstall under Win7, one time it happened with a normal powerdown during the night.
Simultaneously I also investigated my power issues, It turned out my PSU was too weak for the amount of disks I wanted to run leading to weird behavior, but the stupid PSU never turned off when it was overloaded.
So I thought either the power or the old SSD is broken in HW, so I replaced the SSD. However this time I bought Samsung 860 EVO, which is a bit better than 850 (DDR4, newer controller, etc), because a few months ago back I got 850 EVO which I put a fresh installation Windows 10 on, but I wanted the old Win7 to be on the older 850 EVO, so I went into a big chain of cloning and moving. First time I did something like this in this scale, but I have some good experience in general how tackle new things and get the right approach, keep things logged in mind and not rush, but my troubleshooting skills with PCs in general come for help as well, so it not hard at all, it's just takes time and juggling things back and forth.
I probably rebooted the PC like 30 times and that's not the final number, heh. I started with making some image backups, then I directly used clone-disk-to-disk to clone Win10 from 850 EVO to 860 EVO. I don't think I disabled ntfsfix, I don't know about -e1 auto (adjust ntfs geometry) but I did disable resizing and I used "USE PARTITION TABLE FROM SOURCE". It was a first time success, I did some offline full chkdsk tests using the Win10 from 850EVO (had to temporairly change disk MBR ID so it would get online, then change it back, diskpart: uniqueid disk ID=8numHEX)
Win10 Seems to be running just fine, with no issues with any programs whatsoever so far, did 3 cold boots, and that's where I'm typing this message from right now.
It's a different situation with win7 completely ... note that I have since got a new PSU, and I didn't even plug in all the HDDs while doing this.
Motherboard: ASUS P9X79 (from 2012 I think, bought in 2013, it has UEFI, I run CSM and keep avoiding using most UEFI stuff to be on the safe side, so it's all MBR except data disks)
Clonezilla Version: stable - 2.5.5-38
OLD WIN7 SSD: Samsung 840 PRO 128GB
NEW WIN7 SSD: Samsung 850 EVO 250GB
WIN10 SSD: Samsung 860 EVO 250GB
The first time I cloned, with -k1 the cloned thing had the same problem after a second reboot, 0xAA55, in the BIOS this is reported as "NO OS WAS FOUND ..." had to go to TestDisk to fix, also bootsector recovered from backup boot sector, still wouldn't work, it kept saying DISK READ ERROR which never happened before on the old SSD 840 after fixing, so something was worse with the cloned situation.
There's many clones thereafter, I'll avoid explaining each one separately.
1. -k1 proportional partition - created a lot of chkdsk errors, programs broken, DLLs
2. -source disk/image partition - clone and image both created less chkdsk errors but programs were still broken, one of the times it was completely unbootable, the win7 autoloaded into recovery mode and reported system32\kdcom.dll was corrupted, same thing continue to happen after chkdsk fixed a bunch of files but not kdcom, so this was a failure again (used without -r but with ntfsfix and -e1 auto)
NOTE: THE FIRST BOOT IS ALWAYS SUCCESSFUL - Win7 STARTS AUTO DRIVER INSTALL FOR THE NEW SSD AND PROMPTS REBOOT NEEDED - AFTER THE SECOND REBOOT ISSUES START TO BE MORE OBVIOUS WITH ERRORS IN EVENT VIEWER, SCHEDULED PROGRAMS NOT STARTING, SHORTCUTS MISSING.
NOTE2: I USED SHA-1 SUM CHECKING AND MD5 FILE CHECKING IN ALL CASES AND ALL CLONES WERE SUCCESSFUL FROM CLONEZILLA SIDE
So it's all unusable, there's some guides on the net I see they suggest creating partitions manually before, so I'll try that, but the same guide also mentions the driver install for the disk so that appears to be normal.
Next I will try to create partitions manually ahead (using Win10) matching the original size and only clone data, I will also disable -r and ntfsfix and -e1 auto (is this ntfsfix?) - then I will manually resize the partition once it is cloned and checked if it works fine, that's one of the ideas right now.
I only noticed ntfsfix while looking at the fast scrolling verbose output in clonezilla otherwise I wouldn't recalled this thing is ran by default - I haven't figured out how to copy the full verbose clonezilla log from the RAMDISK to the USB, this has been the only big disappointment so far that this is not easier and basically found zero info on the web how to actually view it, but my debian and linux console skills are practically zero also.
After that, if it won't work, I'll go back to 840 SSD and disable all driver installation and see maybe if the driver install has some side effect that produces the NTFS corruption.
Even if the correct way is just around the corner for this Win7 Clone, it's not a problem, I do like to write such procedures in threads not only because it may help others but I find myself months or years later needing the same thing, it's also another backup to my notes which can get lost in the sea of text files.
Last edit: Anonymous 2018-04-19
Well, it's tough, I did the manual thing, recovered from image, it was successful in clonezilla, but didn't boot some 0x000007c error code, failure again, getting desperate ...
For some reason I didn't try ntfsclone or the other options, only partclone.
I just skipped to linux brainstorming, did a dd dev zero on the target drive, and just dd the 128GB onto the 250GB disk. First boot perfect, no ntfs errors, all programs started ok.
Second reboot, corruption prevented some programs from fully loading assets, photoshop that just worked, displayed an error about an unrecognized file, ntfs corruption events, please run chkdsk, and after that it's the same train with chkdsk fixing some 1000 files during boot, still the whole thing remains practically unusable as leftover corruption persists even if there's no corruption events.
I'm beginning to think it's some kind of software thing, either it's the side effect of the driver install for the new SSD or that the whole thing is broken because it's an old install and some drivers or programs could be messing something. But I have one of the toughest maintenances, there's no bloatware, this is not an average hollywood-style teenager type of installation with things thrown all over the place. But I do tweak a lot as well, all of my OSes are nowhere near default settings.
This more looks like it's going to have to be a fresh reinstall on the new SSD.
If I could only figure a way to do some logging to see what the heck is the driver installation doing, I can turn it off probably easily, on Win10 I've done tweaking as well, even heavier in some cases, but it's a much fresher installation but also more careful tweaking that is not too speculative, and it looks like I have done some thing to the auto-driver installation that I keep getting event errors about not being able to install driver for each of the partitions, I didn't even know I kept getting those errors for the 2 months it's been installe, maybe that made Win10 work so this thing should get investigated, I didn't really purposelly disable auto-driver install, I think it was just part of all the other tweaking and I forgot what was the reasoning, probably because it had to do something with windows update.
Last edit: Anonymous 2018-04-19
Well, I decided to do a clean re-install on the 250GB one, considering I just did Win10, I guess I have more fresh memory with windows tweaking and things may go faster than usual, probably justified at this point with a really old setup. Usually I'm determined to get to the bottom, but in my case it's not so critical, I can still keep using the old Win7 install alongside anyway until the new install is finely tuned up.
I could try preventing the driver install, but it's frustrating at this point, i've done full cleans on the SSD a number of times and don't feel doing it anymore.
Even if would be successful, it would be mixed feeling, I would do many reinstalls to move some programs onto the SSD fully since it's now all split to a HDD and messy with hardlinks.
If someone hits a similar scenario, the first thing to do would be to disable the auto driver install prior to cloning since after that things go south, or it has to do something with these specific SSDs aren't the similar enough under the hood.
Last edit: Anonymous 2018-04-22