Files missing during file transfer / renaming process

Support
27

prints line

blacklist ntfs3

and, using sudo privileges, sends it to file /etc/modprobe.d/disable-ntfs3.conf
(creating it if it doesnt exist)
(( also - the ending file will contain only that line … overwriting any contents … the -a flag would have appended ))

What a blacklist does is pretty self-explanatory. It blacklists that module. So if something during start or other operation of the system wants to try and fire up the ntfs3 module … it should fail to do so.

No. It stays this way until you remove the file.

Possible it is. Maybe even likely, skewing more likely as time and abandon wear on.

Delete the file and reboot.

2 Likes
28

@cscs
Thanks for the detailed explanation :slight_smile:

If NTFS3 is really this faulty, maybe @philm should warn his community and maybe Pin this Topic under known issues.

This issue sounds like a big deal to me. :frowning:

1 Like
29

It’s not Manjaro, it’s LINUX (or rather microsoft keeping the source code for ntfs closed).
You are asking Manjaro to list things that are not 100% “supported” on linux, that list would be veeeery long.

30

hrowing my 2 cents… again

just because ntfs-3g works with no complaints, dont automatically assume it is superior to ntfs3. i have found many times where ntfs-3g happily working over NTFS partitions, when there were data corruptions detected only in windows utils (i.e. chkdsk). while YMMV with ntfs3, it will atleast reject mounting corrupted NTFS drives while ntfs-3g does not and makes things worse.

if you value you data, and need interoperability (windows/linux/macos) resort to exFAT as already suggested.

31

I dun think the word “superior” is applicable here.
The right word, imho, should be “stable”.

I’m not familiar with the use of exFAT in HDDs.
For 20TB HDD, would exFAT be recommended?

32

Size of drive doesn’t matter.

From the wiki:

Scalability to large disk sizes: about 128 PB

I think you are ok… xD

33

Out of curiosity, I copied out the specs of NTFS and exFAT from Wiki…

File system exFAT NTFS
Stores file owner No Yes
POSIX file permissions No Yes[ak]
Creation timestamps Yes Yes
Last access/ read timestamps Yes Yes
Last metadata change timestamps No Yes
Last archive timestamps No No
Access control lists No Yes
Security/ MAC labels No Yes[al]
Extended attributes/ Alternate data streams/ forks No Yes
Metadata checksum/ ECC No No
Hard links No Yes
Symbolic links No Yes[bf]
Block journaling No No[bg]
Metadata-only journaling Partial (with TexFAT only) Yes[bg] (2000)
Case-sensitive No Yes[bh]
Case-preserving Yes Yes
File Change Log No Yes
XIP No ?
Resident files (inline data) No Yes (approximately 700 bytes)
Internal snapshotting / branching No No
Encryption No Yes
Deduplication No Yes[cb][55]
Data checksum/ ECC No No
Persistent Cache No No
Multiple Devices No No
Compression No Yes
Self-healing[ca] No No
Host OS misc. Windows
Offline grow No Yes
Online grow No Yes
Offline shrink No Yes
Online shrink No Yes
Add and remove physical volumes No No
Sparse files No Yes
Block suballocation No Partial
Tail packing No No
Extents Partial (only if the file fits into one contiguous block range) Yes
Variable file block size[cj] No No
Allocate-on-flush No No
Copy on write No ?
Trim support Yes (Linux) Yes (NT 6.1+; Linux)
34

Who cares if its Manjaro or not. Linux users may run into big issues when they using the new NTFS3 and this shouldnt be hidden or you want to see more users running into this trap?

This is the second Topic related to NTFS3 (im not even searched the Forum) data corruption and these people facing really big problems now from it.

I remember the nvidia driver issue around 3 years ago which forced all nvidia GPU owners to reinstall the driver (470.xx if i remember right) though tty after a stable update, there was a big message in this Forum and this nvidia driver was also closed source.

1 Like
35

I just can say, that on kernel 6.1 ntfs3 is included, but not used by default. I have not tested it, but I guess those users use a stable or mainline kernel, which enables ntfs3 by default.

For example, if I use udisks2, as any file manager, it use ntfs-3g (fuseblk) and not ntfs3.

udisksctl mount --block-device /dev/DeviceName
mount -t ntfs3
mount -t fuseblk

Furthermore, stable kernels are considered to be stable, but in fact still beta, not to mentions the mainline kernel, which is for development. That should be clear to users, who switch to higher kernel versions.

1 Like
36

well, it is still happening…

37

Well. Tring to install using tty but still not done.

38

Well duh.

NTFS AND Nvidia are BOTH closed source, so OFC they can both act “buggy”.

The option is no NOT use either ntfs or nvidia, complaining about “manjaro not informing us” is… idk what to say.
This is nothing new.

And that a 40-series gpu is not yet up to par as nvidia gpus that has been in use for years, what a surprise! (sarcastic)

I am AWARE that nvidia is problematic, ntfs as well. So with ntfs I have precautions taken (in my case backups with urbackup) OR I WOULD STOP USING IT AND CHANGING FILESYSTEM!

39

I have a few questions about the ntfs3 issue and how to evade it.

1.Do i really have to blacklist ntfs3 or is it still safe to use LTSC Kernels like 5.15 or 6.1?

2.Im using exfat for my backup drives but sometimes i backup few of my games/savegames from windows, can ntfs3 still corrupted my files when i mount my NTFS Partition from Windows in dolphin to backup this games to my Exfat Partition?

3.What happends if i use it vise versa and copy my games from exfat to ntfs, while im on Linux?

40

If your question is: Do I need to blacklist ntfs3 if you only want to use ntfs-3g. Then the answer is no, you mount it with ntfs-3g and that:s it.

not if you make sure that it is mounted with ntfs-3g.
It seems that above posts indicate that if you mount it in dolphin it will use ntfs-3g, correct me if I’m wrong.

Same thing, all depends on how you mount AND the fact that I hope you have become aware that NTFS IS NOT GREAT TO USE ON LINUX!

41

if i want to make sure that i only mount it with ntfs-3g, i wouldnt need blacklist also… sorry but your answer dont help me at all.

And yes i want to mount my ntfs partitions with dolphin… as i wrote that.

42

Sure, if you rather go the way of getting errors and THEN changing it instead of actually making sure you mount it correctly to begin.

cat /proc/mounts gives you all the info on how all mounts are done on your system, could be good to know.

There is no “right or wrong way” here, you do what you feel is best! :+1:

43

Just add the driver to blacklist and you are good.

The ntfs3 module is the open source version of this one: Microsoft NTFS for Linux by paragon software - Introduction You get support and a license for ~40 €. It is partially written in C++ and for the kernel it has to be plain C, therefore an amount of functions has to be rewritten from scratch. The commercial version should work good (not tested), but the open source version is still not mature. I guess chkntfs (chkdsk under windows) and mkntfs will stay in the commercial version.

44

I want to get sure, thats why i asked here in this forum.

Strange, when i mount my ExFat Partition in Veracrypt it only shows me (Fuseblk):

cat /proc/mounts:

veracrypt /tmp/.veracrypt_aux_mnt1 fuse.veracrypt rw,nosuid,nodev,relatime,user_id=0,group_id=0,allow_other 0 0
/dev/mapper/veracrypt1 /media/veracrypt1 fuseblk rw,nosuid,nodev,relatime,user_id=0,group_id=0,default_permissions,allow_oth
er,blksize=4096 0 0

I though that fuseblk used for NTFS but also for exfat?

So all in all i should be fine as long as cat /proc/mounts telling me it used fuseblk?

45

And I tried to explain that in MY humble opinion its better to make sure you have full control of the system, in this case mounting how YOU want to mount rather than disabling something that (you think) is being used.
You do what YOU want, not sure how to explain my stance more clearly. :frowning:
For you maybe it feels better to just blacklist it, then do that and if you get errors, deal with them then. :slight_smile:
I don’t think there is a correct and incorrect way here, its YOUR way!

Not sure how that works to be honest, but would not be surprised if fuseblk is used for most mounting. I VERY rarely mount with dolphin, I have configurations on how I mount, I prefer systemd over fstab.

But your answer prompted a reaction from me. I checked my mounts and you are correct. The ntfs is using fuseblk to mount my ntfs and doesn’t tell me what module, but if I check my systemd mount that I’m using to mount it tells me I’m using ntfs-3g.
I guess you have to find out HOW fuseblk is mounting. I SUSPECT, I do not know, but suspect it is still using old ntfs-3g and fuseblk2 is using ntfs3, but I can be completely wrong here.

$ cat /proc/mounts | grep /dev/sdb2
/dev/sdb2 /media/fil-fitt fuseblk rw,nosuid,nodev,noexec,noatime,user_id=0,group_id=0,default_permissions,allow_other,blksize=4096 0 0

$ systemctl status media-fil\\x2dfitt.mount
● media-fil\x2dfitt.mount - File drive (fil-fitt)
     Loaded: loaded (/etc/systemd/system/media-fil\x2dfitt.mount; enabled; preset: disabled)
     Active: active (mounted) since Tue 2023-08-01 15:29:16 CEST; 2h 43min ago
      Where: /media/fil-fitt
       What: /dev/sdb2
      Tasks: 1 (limit: 37331)
     Memory: 386.5M
        CPU: 7.010s
     CGroup: /system.slice/media-fil\x2dfitt.mount
             └─724 /usr/bin/mount.ntfs-3g /dev/sdb2 /media/fil-fitt -o rw,users,noexec,nosuid,nodev,noatime,uid=1000,gid=1000,dmask=0022,fmask=0133

Aug 01 15:29:15 bednaManjaro systemd[1]: Mounting File drive (fil-fitt)...
Aug 01 15:29:16 bednaManjaro ntfs-3g[724]: Version 2022.10.3 external FUSE 29
Aug 01 15:29:16 bednaManjaro ntfs-3g[724]: Mounted /dev/sdb2 (Read-Write, label "Lilla Fil-fitt disken", NTFS 3.1)
Aug 01 15:29:16 bednaManjaro ntfs-3g[724]: Cmdline options: rw,users,noexec,nosuid,nodev,noatime,uid=1000,gid=1000,dmask=0022,fmask=0133
Aug 01 15:29:16 bednaManjaro ntfs-3g[724]: Mount options: users,noexec,nosuid,nodev,allow_other,nonempty,noatime,rw,default_permissions,fsname=/dev/sdb2,blkdev,blksize=4096
Aug 01 15:29:16 bednaManjaro ntfs-3g[724]: Global ownership and permissions enforced, configuration type 7
Aug 01 15:29:16 bednaManjaro systemd[1]: Mounted File drive (fil-fitt).

And if anybody cares, here is my mount unit file:

[Unit]
Description=File drive (fil-fitt)

[Mount]
What=/dev/disk/by-uuid/CE8C0EC38C0EA655
Where=/media/fil-fitt
Type=ntfs-3g
Options=users,auto,nofail,noatime,uid=bedna,gid=bedna,dmask=0022,fmask=0133
#DirectoryMode=0755
TimeoutSec=15

[Install]
WantedBy=multi-user.target

Edit
For those of you that are scandinavian, I AM SO SORRY for my childish name of the drive, it has been following me since I was a teenager and built my first comuter in the late 90:ies. :flushed:

46

fuseblk is a placeholder for any filesystem which runs in userspace. So extfat and ntfs-3g are userspace drivers. So anything what uses fuse, is user space.

ntfs3 is a kernel space driver, just like btrfs or ext2/3/4, also exfat can be installed as kernel space driver with a dkms.

2 Likes