Fat32 partition size limit (OT in uk.d-i-y)

[Mayayana]

| > What restrictions are you referring to? Linux supports | > around 19 different file systems. | > | Mayayana is referring to the file permissions model (755, 644 etc). | Yes. Using Linux partions -- ext* -- for data can result in complications, so I just use FAT32.

| And EXT2 on Linux, would support file permissions in similar | ways to NTFS under Windows. The difference is, the permissions | model for NTFS just never "clicked" with me, leaving me permanently | unable to explain it thoroughly to others. I can only figure out | the simplest of things on NTFS. (Like the read-only folder bit | being overloaded as a folder customization flag.) If you read | descriptions of this stuff up to the domain level, you're left | speechless.

The API code to deal with it is also "not to be believed". :) I remember once seeing a humorous comparison of the steps required in Windows vs Linux. Security through abstruseness.

[J. P. Gilliver (John)]

It may work better in the handling of sectors, and possibly the journalling (I'm still not convinced about those); it's the permissions aspect that really bugs some of us. Now, if there was a protocol that Windows understood that had the (claimed) advantages of NTFS, or it were possible to use NTFS but turn permission control off, ... but as another has said, you can't )-:. []

[Huge]

Jesus, but that's dumb.

[Roger Mills]

My OP concerns a large back-up disk. My rationale for splitting it into a number of partitions is to reduce the risk of losing *all* my data if one partition gets corrupted. I don't know whether that's valid or not.

[Mayayana]

| > Yes. Using Linux partions -- ext* -- for data can | > result in complications, so I just use FAT32. | | Jesus, but that's dumb. |

In 4 words you've managed to swear and use dubious syntax, all while not actually saying anything. Care to try again? Why not go for broke and write a complete sentence or two. That is, if you can spare the time. :)

If you mean that ext* is superior to FAT32, it's not in my usage. I'm not doing a lot of intensive file operations. But ext* can cause problems if I save a file while logged on as one user and then try to access it again later while logged on as another. It's built-in "security" that can't be turned off. Ditto for NTFS. Thus, FAT32 is a good solution for both Linux and Windows.

If you mean it's dumb to have data partitions, I can only say that it works for my backup needs. I see no reason to allow data to share in the risk of a possible unbootable OS. That risk simply isn't necessary. To put everything in one partition is like having a tractor trailer truck that doesn't allow the cab to be separated from the trailer, so that something as simple as a leaky radiator could end up spoiling an entire load of perishable product.

[Gene E. Bloch]

I find it easier to believe a whole disk would die than that a single partition would, but I don't know enough to bet money on that.

However, my favored policy is to back up alternately to two different drives, and have each connected only while its backup is in process.

[Gene E. Bloch]

You say that, but have you ever had a disk that has a leaky radiator? You really shouldn't talk unless you've actually had that experience.

;-)

[Huge]

FAT32 isn't a good solution for anything. Other than people too stupid to work out how file protections work.

[Ken Blake]

Ditto!

[pjp]

The concern is basically does disk itself go toast. I've never had a partition go south except for if the hard disk is failing.

I also see no reason for multipule partitions any longer. Find you must have them then buy another hard disk, they're cheap enough per Gb.

[Brian Gregory]

That's what I usually do.

Then I change the security on the partition so 'Everybody' has 'Full Control' to make sure I'll always have write access even if my user name isn't the same on all systems.

Lately I've also been choosing a larger allocation block size than the default 4K, hoping that this'll speed things up in some cases (I use 32K).

[Brian Gregory]

It's a totally artificial limit introduced fairly recently my Microsoft. My theory is that they're hoping to push people into using their new proprietary exFAT format.

I thought the command line format command could still do FAT32 >32GB but maybe I'm wrong.

[John Rumm]

No you don't, he was wrong anyway (quelle surprise)

(Fragmentation of free space does not have an impact on the total free space or indeed the size of file that can be written, however it does mean that a new files may end up being fragmented from day one if there are insufficient contiguous blocks available to hold them)

Sticking all user data on a separate partition makes it nice and easy to backup all your stuff with a whole partition backup without needing to backup all your apps and OS each time with it)

[Brian Gregory]

It couldn't be easier to get access to what you want on Linux -- just sudo chown or sudo chmod.

Anyway much of the stuff you don't have access to is for a reason -- if you accidentally mess with it you'll break stuff.

With Windows it is more complicated but if you just store your data on a partition where everything inherits access where 'Everybody' has 'Full Control' there's no problem. Or just learn how to claim ownership.

[polygonum]

Rather depends on what you are doing!

Space management is obviously a reduced overhead when considered across the whole partition, but the overheads of accessing a single file? Probably not much.

Could end up costing a lot of space if you have a lot of files that are small.

I certainly do things like that when creating partitions for backups and other very large files.

[Andrew Gabriel]

Another factor is the simplicity. This was very important with the early digital cameras and phones, where you can fully implement a FAT filesystem in only 2kB of memory. If you only need read access, that can be done in less than half a kilobyte.

[J. P. Gilliver (John)]

[]

In hardware terms, I suspect you're right. However, in software terms, it _is_ possible to screw up a partition. Although this is more likely to happen with a system partition than a data one. I'd say there is - arguably! - validity in partitioning your working disc into system and data (apart from anything else, the image/clone of your _system_ then doesn't take so long to make, so you might do it more often). For a disc used entirely for backup purposes, there's probably less reason to partition it.

Sounds good policy.

[J. P. Gilliver (John)]

In message , Rod Speed writes: []

[]

[] Not entirely. The reasons JBG give may or may not be valid, but to say there is _never_ any point to it can't be correct.

[J. P. Gilliver (John)]

There's a difference between not being able to work it out, and not seeing the necessity to bother. For most home users, the only advantage of NTFS is the larger filesize limit, and the advantage of FAT32 is that more (especially older) devices can use it. With the exception of large video files (which are probably becoming commoner now), I suspect few people will be bothered by the filesize limit; similarly, I suspect few people will come across equipment that can only recognise FAT32 these days.

[J. P. Gilliver (John)]

Is it just the filesize limit that makes it worth doing all that (i. e. if you used FAT32 you wouldn't have to bother), or are there other advantages you see to NTFS?

I guess the amount of effort is probably about the same, for someone with an XP (or is it W2k?) or later Windows system: to make a large FAT32 partition (which is natively universally accessible), you have to remember where you put your big-FAT32-formatting utility; to make an NTFS one, you have to mess about with the permissions as you describe above. If I'm understanding you correctly, you only have to do it once whichever system you go for.

(Have you found any difference? [No axe to grind - just curious.])