Several questions on installing Tumbleweed along with Win10 booting with UEFI

[cnzhx]

Hi all,

I am going to ask a lot questions here because I will install the openSUSE Tumbleweed along with Windows 10 in a new laptop. This is a new situation for me because I think my knowledge on this kind of things are still in the stage of 5 years ago, and I am also a relative new user to Linux. Please forgive me if my questions seem silly to you.

Usually I would ask one question a thread, but for this kind of things, I think it’s better to group them together because lot of information is correlative. Here it the basic hardware information:

• CPU: Intel i7-7600U
• Video card: Integrated GPU
• Drive: 512GB SSD (M.2 2280 NVMe PCI-e)
• Memory: 8-24GB RAM (I will increase the original RAM of 8GB to a larger size later)

Here comes the questions.

1. Swap size and swap partition
I am not going to use hibernation any more given the high read speed a SSD can provide.

I read that for the situation of a large size of physical memory, it’s better to keep a small to zero swap if no intensive task will be involved. Then my question is, for zero swap, should I just do not assign a swap partition? Would this be the right/safe recommendation?

If it is safer to use a small swap partition than no swap, may I put this partition behind all other partitions in case of easy to increase its size if I would need it? And I already know that I should change swappiness to a lower value such as 10 if there are plenty of DRAM.

2. Settings related to SSD
This is the only storage drive in this laptop.

As for SSD, are there any settings apart from the default ones provided in openSUSE Tumbleweed I should consider to keep the SSD in good performance and state?

And yes, I know that SSD is for using not being caring. I just want to reduce unnecessary wearing.

3. File system types
I planned to use Btrfs for root (/) partition because it is the default one. But I do not know should I use it for /home as well. Any recommendation?

These are all that I am trying to figure out now. But most of them seem to be opinion based rather than technical questions. So please give me your experience-based opinions. They will be very much appreciated.

Regards,
CnZhx

[gogalthorp]

You should have some swap just in case. Don’t need as much as is if you hibernate

If using BTRFS there are none that are required

Default is XFS which is fine or use ext4. There is little difference in FS any more

[cnzhx]

@gogalthorp

You should have some swap just in case. Don’t need as much as is if you hibernate

If using BTRFS there are none that are required

Default is XFS which is fine or use ext4. There is little difference in FS any more

Thank you gogalthorp for your advices.

I will use a small swap partition then. Maybe 1 or 2GB is a good start. Do you think it’s a good practice to make it as the last partition on the drive?

[gogalthorp]

Makes little difference these days with HD and none with SSD.

[cnzhx]

@gogalthorp

Makes little difference these days with HD and none with SSD.

Make sense

I am still thinking this like on a spin drive. But it’s true that I am not aware of the mechanism behind, such as how does the SSD store the contents. Is it pseudo random, or something like this, unlike the sequence from lower to higher in a spin drive?

[gogalthorp]

It is random access memory just made to look like a drive organization.

It uses flash memory which has odd characteristics. The memory is setup in blocks which are mapped to the virtual sectors on the virtual drive the OS sees. Flash can not be changed it has to be erased a block at a time then it can store new data again. So any writes even a singly byte means that a block is moved to a recovery area (virtually not physically) and a new unused block is mapped in it’s place and can be written. The hardware then will erase the used block that was set aside at a convenient time and it is added back into the free block pool. Additional memory is provided to act as a pool. Flash has a finite number of erasers that can happen before it dies. But the hardware has what is known as write leveling procedures which assure erases are done balanced across all memory thus increasing the life of the device. Modern drives can sustain 100s of terabytes of writes and generally are rated at about 5 years normal usage before exhausting the extra memory pool and begin to deteriorate.

I use the following mount parameters in /etc/fstab for a ext4 FS

ext4       noatime,acl,user_xattr 1 1

noattime stops writing the date-time a file is accessed which happens often and reduces the life of a SSD

So more then you wanted to know???

[nrickert]

@cnzhx

I read that for the situation of a large size of physical memory, it’s better to keep a small to zero swap if no intensive task will be involved.

I don’t think it matters. I have a lot of swap, but it is almost never used. And since it isn’t used, I’m doubting that there’s any “intensive task” happening.

[cnzhx]

@gogalthorp

It is random access memory just made to look like a drive organization.

It uses flash memory which has odd characteristics. The memory is setup in blocks which are mapped to the virtual sectors on the virtual drive the OS sees. Flash can not be changed it has to be erased a block at a time then it can store new data again. So any writes even a singly byte means that a block is moved to a recovery area (virtually not physically) and a new unused block is mapped in it’s place and can be written. The hardware then will erase the used block that was set aside at a convenient time and it is added back into the free block pool. Additional memory is provided to act as a pool. Flash has a finite number of erasers that can happen before it dies. But the hardware has what is known as write leveling procedures which assure erases are done balanced across all memory thus increasing the life of the device. Modern drives can sustain 100s of terabytes of writes and generally are rated at about 5 years normal usage before exhausting the extra memory pool and begin to deteriorate.

Thank you very much for the explanation. It’s clearer and easier to understand than the articles I have read. Maybe it’s because your words are tailored for my needs

@gogalthorp

I use the following mount parameters in /etc/fstab for a ext4 FS

ext4       noatime,acl,user_xattr 1 1

noattime stops writing the date-time a file is accessed which happens often and reduces the life of a SSD

So more then you wanted to know???

I do have a question on this. Does “access” here mean “read” only? I do want to know the create and modify date. If no access time is written, maybe I cannot sync files by timestamp?

@nrickert

I don’t think it matters. I have a lot of swap, but it is almost never used. And since it isn’t used, I’m doubting that there’s any “intensive task” happening.

Hi nrickert,
Yes, I am not sure if intensive tasks will be involved in my operation in future. But there is definitely no intensive tasks for at least half a year for me because I am busy in writing my thesis.

So, if you actually do not need swap, why do you keep a lot of swap? Do not you think it’s a waste of drive space?

[gogalthorp]

Access fo a file means you open it. Normally the date-time of the files is updated at that point which causes a write to the disk. Since the last time the file was opened/accessed is normally not important turning off the function with noatime saves a write and thus ware on the flash memory is reduced. This is not needed by BTRFS but is for ext4.

If you want to see all the mount options type man mount into a terminal. note man is short for manual and most all Linux commands have an entry

Swap is one of those thing that may or may not be needed depending on how you use the computer. If you run out of main memory the OS just barfs.

[nrickert]

@cnzhx

So, if you actually do not need swap, why do you keep a lot of swap? Do not you think it’s a waste of drive space?

I have more drive space than I will ever use. So throwing a bit into swap doesn’t hurt.

[Fraser_Bell]

@gogalthorp

So more then you wanted to know???

au contraire … very worthwhile info, g.

[cnzhx]

@gogalthorp

Access fo a file means you open it. Normally the date-time of the files is updated at that point which causes a write to the disk. Since the last time the file was opened/accessed is normally not important turning off the function with noatime saves a write and thus ware on the flash memory is reduced. This is not needed by BTRFS but is for ext4.

Thanks a lot. I may also not need access time to be recorded.

One thing more: so the BTRFS’s feature is snapshot, which is not needed by /home, at least for me. Then the recommended file system for /home is EXT4 or XFS. Am I right?

After a little more digging, I think EXT4 is better than XFS for /home for me. But how about BTRFS?

@nrickert

I have more drive space than I will ever use. So throwing a bit into swap doesn’t hurt.

Good point

[cnzhx]

@cnzhx

After a little more digging, I think EXT4 is better than XFS for /home for me. But how about BTRFS?

Sorry. I meant to say “how about BTRFS with snapshot disabled”.

[cnzhx]

@cnzhx

Sorry. I meant to say “how about BTRFS with snapshot disabled”.

After a thorough reading on the discuss here <https://forums.opensuse.org/showthread.php/502082-BtrFS-vs-XFS-vs-Ext4> and the test results on <http://www.phoronix.com/scan.php?page=article&item=linux-410-earlyfs>, I decided to use BtrFS for / with snapshot enabled, and EXT4 for /home and /data because there are plenty of tools to rescue it in case of a surprise.

Thank you all for your advice and attention