Swapfile- what disk filesystem?

[Barkingmad]

Hello

If I want/need to set a swapfile does it need to be on a Linux formatted partition eg ext3, or can it be put on ntfs?

I'm currently trying Puppy on an old laptop and there are several partitions from Dell's system recovery stuff so it would be nice to avoid trying to resize things.

Thanks for any help
Will

[backi]

@Barkingmad

Hi !
You have to make a (separate) Swap Partition and format it with Gparted (same as you do when creating a Partition with a specific Format) >just Format to>linux-swap -----(instead of ext2/3/4 or fat32 etc).
linux-swap is another Kind of Partition Format like ext2/3/4 or fat32 etc in its own.

[Feek]

Hello

If I want/need to set a swapfile does it need to be on a Linux formatted partition eg ext3, or can it be put on ntfs?

Read this instructional how-to:
viewtopic.php?t=3933

I don't use swap file but it seems you can place it anywhere (personally, I would place it on the fastest drive available).

Be aware of this warning:
"If Windows is on the internal drive.
Run Windows defrag program, to defrag the partition, before putting the pup swap file, on it."

[backi]

@Barkingmad
Sorry ....did not read your Question well......you were asking for a swapfile .....not a swap Partition.......is it right?

[Wiz57]

I use several different Pups, mainly ScPup32 20.05, though also ArchPup 32...anyhow, all my Pups are "installed" into
their own directory ("folder") on the internal HDD which is NTFS formatted, contains Windows XP Home. I created a
swapfile in the "root" directory of my C: drive, using the tools included in ScPup and other Puppies (called Pupswap).
So yes, you can use NTFS formatted drive without major problem. The primary issue I have noticed is not with the
swapfile itself, but something in the way series 5.X Linux kernels handle NTFS...maybe something changed in the
ntfs3g "driver", I don't know, but so far any series 5.X kernel results in super slow access of ANYTHING on an NTFS
drive. Even opening a file manager, such as RoxFiler or PCManFM takes super long time...this also shows up whenever
any access of the C:\PUPSWAP file is attempted by the system, most notably when web browsing (my old Acer Aspire
netbook only has 1 gig RAM)...anyhow, downgrading back to older 4.169 or so kernel along with matching fdrv "fixes"
this issue.
Wiz

[Barkingmad]

Thanks to all

Wiz57 -- That's pretty much the setup I've got so I should be good to go though I'll take the tip to boot Windows and run defrag straight before.

Bye
Will

[Barkingmad]

Hello -- to follow up

I've now had chance to try this. Used Windows defrag then rebooted to Puppy, used the Pupswap tool to make a swapfile then made a savefile on shutdown. A 'cold' reboots finds and uses both so all seems well.

Also, thanks for the heads up about the Linux 5 series kernels and NTFS, I'll avoid them for now on duel boot machines.

Bye for now
Will

[Phoenix]

Any filesystem will do as long as it doesn't incur significant I/O (input/output) delays. However note that for NTFS you may end up possibly corrupting it if you forcibly shut off the computer while its running an OS that is not Windows. As well defragmentation is not necessarily required. For flash-based drives (SSD, USB, SD cards, etc) there is little to insignificant penalty for fragmentation. (Unless you have lots and lots of very very tiny pieces scattered around) And its best not to since flash-based drives can only take so many writes before going down. For HDD, hard disk drives it does matter because the disk platter needs to be spun around to access the location. So if its fragmented it will be spinning all over the place.

Although there is one case where you may end up having to defragment, and that is when filesystems have limitations about it. NTFS can only keep so many scattered pieces in a record before it can't keep track and stops you from writing anything. This in the very early days of computing could confuse users as their disk wasn't full! But their disks had extensive fragmentation. EXT filesystems do not have any fragmentation limit I am aware of.