NetBSD 6.0 Is On Approach With New Features

Quote:

---

Originally Posted by vertexSymphony

Which filesystems are you comparing ?

---

ext2 (async) vs ufs (sync)
ext4 vs ufs

Quote:

---

Originally Posted by LightBit

ext2 (async) vs ufs (sync)
ext4 vs ufs

---

UFS (I assume UFS2?) with which features enabled? (softupdates, journaling,etc ) ...comparing UFS2 against ext2 is unfair and shouldn't even be mentioned.
And sorry, but UFS, even with journaling is not rock-solid ... if you really need that robustness (even if it's weird, because in practically everywhere you usually have UPS units), ZFS might be a better choice, but this is a different beast on his own and should be compared to a similar like btrfs.

Quote:

---

Originally Posted by vertexSymphony

UFS (I assume UFS2?) with which features enabled? (softupdates, journaling,etc ) ...comparing UFS2 against ext2 is unfair and shouldn't even be mentioned.
And sorry, but UFS, even with journaling is not rock-solid ... if you really need that robustness (even if it's weird, because in practically everywhere you usually have UPS units), ZFS might be a better choice, but this is a different beast on his own and should be compared to a similar like btrfs.

---

There are no reliability differences between UFS1 and UFS2 (without additional features).
With no features (like softupdates, journaling) enabled it's more reliable. Sync file systems are far more reliable.
UFS is still more solid than ext*. And there is also hammerfs.
Why comparing UFS2 (without softupdates or journaling) against ext2 is unfair?

Quote:

---

Originally Posted by LightBit

ext2 is asynchronous by default -> not reliable
Data loss on power failure or crash with ext4, especially early versions (like 2.6.30).

Compare /etc.

---

Data loss on BSD file systems as well. On Linux you have much better choice.

Quote:

---

Originally Posted by LightBit

This is not proof.
*BSD might be better server than Linux, because it's more careful with your data ... It doesn't risk for better performance.

---

It's not more careful. Linux also doesn't risk stability for performance, but it's a matter of configuration. You can use Ext3/4 with safe options and you're done.

Quote:

---

Originally Posted by kraftman

Data loss on BSD file systems as well. On Linux you have much better choice.

---

Data loss on BSD file systems is less likely than on Linux. I had corrupted ext2 (fsck failed to fix it) and lost data on ext4.
What better choice? btrfs?

Quote:

---

Originally Posted by kraftman

It's not more careful. Linux also doesn't risk stability for performance, but it's a matter of configuration. You can use Ext3/4 with safe options and you're done.

---

Only truly safe option is sync, which is not available for ext3/4.

Quote:

---

Originally Posted by LightBit

Only truly safe option is sync, which is not available for ext3/4.

---

There are options for immediate write on both ext3 and 4. Not only that, but in every configuration I have ever known ext4 has a huge performance advantage over any version of UFS safe options or not.

Quote:

---

Originally Posted by LightBit

Data loss on BSD file systems is less likely than on Linux. I had corrupted ext2 (fsck failed to fix it) and lost data on ext4.
What better choice? btrfs?

---

No, it's not more likely on Linux. BSD file systems suck and ext2 sucks as well... Use Ext 3 or 4 with safe mount options. There is sync in ext4...

Quote:

---

Originally Posted by kraftman

No, it's not more likely on Linux. BSD file systems suck and ext2 sucks as well... Use Ext 3 or 4 with safe mount options. There is sync in ext4...

---

100% agree.