That space is reserved for the journal. Basically ntfs or ext3 will “steal” some space overtime when you add files while ext4 reserves it from the beginning
Thanks for the info, one more thing, I have just tried moving a file to the ext4 drive 70gb and it’s taking forever to move any reason why this should be? Could it be faulty drive?
Move from where to where? Using which technology?
I’m trying to move my 4k movie files from 1 external hard drive formatted in ntfs to another external hard drive I’ve just formatted in ext4 both connected to my windows 7 laptop
If both are connected via USB depending on the USB configuration they may share the same USB Port/Bandwidth, making it slow
You formatted it ext4 on windows 7 laptop?
1 Like
In my experience copying usb to usb using one computer is always very slow nothing to do with file systems
Yes using AOMEI partition assistant
Excuse me if I’m being a tad ignorant here, I’m not as technically literate as most posting in this thread but wouldn’t the drive type also affect performance?
For example when I got my first Vero 4K and before I’d worked out using fstab/autofs I had a NTFS formatted, powered 3.5" external hdd plugged into the spare USB port on my Vero and it worked flawlessly even with the infamous John Wick UHD rip,
but if I’d used a 2.5" drive even with a powered hub wouldn’t that make things slower? As I understand it most 2.5" drives run at 5400 RPM rather than the 7200 my powered drive ran at and the 2.5’s seem to always have lower MB’s for the buffer.
A 2.5” drive would still be fast enough
Is the Raspberry Pi 3B+ a similar spec?
This piqued my interest, because sometimes download and extracting to the same USB causes a lockup/lockout until disk activity has finished. I know that I’m probably pushing the limit.
Downloading only reaches a max of 8 to 10MB/s whilst a file is being extracted too (tbh, similar download speed without an extraction).
Running the HDPARM command above gives a “Timing bufferd disk reads” result of 29.46MB/sec.
As far as I can recall, it’s an EXT4 partition.
Not sure I follow the question.
Filesystems are handled the same as they are both Linux platforms.
Sam
You are looking at a few different things here but I think your answer is that for what you are looking at the 3B+ will probably act slightly faster. The network controller changes from a 100mb/s to a gigabit BUT it is still connected the same way. All Pi’s prior to the Pi4 only have a single USB2 connection to the CPU so the ethernet and USB drives still have the same 300mb/s shared total bandwidth.
What I have found is this link regarding when you format a drive in ext4. 5% is reserved for os
See this link How To Free Reserved Space On EXT4 Partitions | Odzangba Kafui Dake
Can someone please tell me if I can do what is done in the link, regarding the command prompt on a windows 7 laptop or if it can only be done on a linus pc
Regarding reducing the 5% reserved for os to 0% as it’s just a hard drive for movies
I have lost 119 GB by formatting in ext4 on a 8tb drive. I have copied all movies from the drive formatted in ntfs to one formatted in ext4 and that’s the difference
Just seems quite a lot of space to loose
Hope you’re familiar how to login via ssh into your OSMC pi/vero:
- Identify the disk’s device name using
dfcommand. Let’s assume you found it assda1. sudo tune2fs -l /dev/sda1 | grep 'Reserved block count'sudo tune2fs -m 0 /dev/sda1- Verify the change with the command in step 2.
Thanks for taking the time to post that
I’m not familiar logging in via ssh
But will give it a go tomorrow I’ll google it and try to find out what I need to now before I try
Many thanks
Details regarding how to access the command line interface can be found here on our Wiki: Accessing the command line - General - OSMC
1 Like
Just to give my 2 cents…
I have heard everyone say that NTFS doesn’t work as well as exFat or ex4 on the Vero 4K (linux).
Everyone talks about overhead and NTFS having to run in userland (FUSE) and what not, but if I’m not mistaken exFAT runs in user land as well and I’ve never seen anyone show any benchmarks.
Honestly, if you have stuttering issues, I doubt it’s the file system.
The differnce in some synthetic benchmark, which I have never seen, would be negligable I’d imagine.
Nothing you’d notice in the real world.
I have a Vero 4K+ with 3 HHDs plugged in via USB all running NTFS.
I have a Seagte Backup Plus Hub 8TB (3.5") which gives you two additional powered USB 3.0 ports.
In those ports I have a Seagate Backup Plus 5TB and 4TB (2.5").
Not once have I ever had issues with stuttering, ever, and I watch a lot of UHD remuxes with Atmos and DTS-HD MA.
If you are never going to plug the HDD in to anything else, then yeah, might as well run ex4.
But then adding files to the HDD will be bottle-necked by the Vero’s USB 2.0 ports
When transfering a 50GB+ UHD remux that can be annoying in my opinion.
But since my HDDs are NTFS, I can just plug them in to my laptop and transfer the files over at 170MB/s+ instead of over the network at <30MB/s.
Now if the Vero had USB3.0 ports instead of USB2.0 ports…
That’s not correct. FUSE will absolutely be the bottleneck here.
Writing to the drive (as some do with torrent clients) will deliver horrendously bad performance. It would be a disservice to suggest that NTFS is a good solution or even acceptable
What’s the actual bottle neck in MB/s?
If it’s higher than USB2.0 than it’s a moot point.
Also, if someone is grabbing torrent by there Vero 4K than their internet speed would have to be higher than this bottle neck too (which is possible).
My 2 cents are just that, my 2 cents.
Everyone has different use cases.
All I’m pointing out is that despite all the fear mongering about NTFS on Vero, I have had no issues other than the USB2.0 bottle neck, Which because I use NTFS, I can easily work around.
That’s great information to know that the OP can consider.
I’ll format a drive as ex4 and transfer a 1GB file to it from the Vero’s local 16GB memory, then I’ll do the same thing with an NTFS formatted drive and we’ll see.
But if I remember correctly, this whole thread was begun because of stuttering and the only stuttering I’ve ever seen is on those super high bitrate jellyfish demos, and those same files stuttered when being played back by the Vero’s local 16GB so… the point is, with real world files I have never experienced any stuttering even with NTFS.
I don’t debate that ex4 is not better, but to claim that stuttering is caused by NTFS… well, I have to disagree with that.
You likely won’t do above 3-4MB/s write. That’s well below that of USB2 and generally below that of most high speed internet connections.
To make matters worse you are also moving your IO bound operations such that they also become CPU bound. If CPU utilisation rises, your IO tanks.
You’re testing from a read perspective not a write perspective.
It’s not fear mongering, it’s fact. I’m not talking about corner cases or high nitrate content. The fact is that sustained throughout via FUSE will always prove challenging.