ext3 or ext4

[hln98]

Hello,

I am at the point where I am going to rebuild my RAID 5 with the addition of the 4th 2TB HD for my TS-439. Since I have very little data on the existing 3 2TB drives, I will drop the RAID and start building a new RAID 5 instead of going to the expanding process which can take a few days. Building a RAID5 with 3 drives last time only took me 9 hours.

My question is I should go with "ext4" instead of "ext3" correct for performance boost in RAID5, correct? I have 2 Macs and one Windows 7 64 bit box. I also read there were some issues with QNAP freezing when accessing files from a "64" Vista machine.

For all the experts, I should go with "EXT4" correct?

I am on the latest firmware 3.1.2

Thanks.

[Q]

hi

with firmware 3.1.2 i would still choose ext3.

with firmware 3.2.0 i think it's safe to use ext4 (there was a bugfix related to ext4, and the kernel was updated). and yes you should get a performance boost.

3.2.0 was released today but pulled back because there was an error with dutch web ui not working. so i guess tomorrow it should be available again.

[hln98]

So with firmware 3.2, I should go with ext4?

[erCicci]

Same for a TS-210?

I have just installed a TS-210 with a couple of Seagate 500GB 7200.12 HDs in RAID1 and I noticed a really poor performance in write operation... transferring a large file to the NAS I gained a transfer rate max of 15MB/s, while in reading from NAS I gained a max of 38MB/s (I have a Gbit LAN, of course ).

Do you think with ext4 I will gain some boost at least in writing operations?

[Complicated]

Do you think with ext4 I will gain some boost at least in writing operations?

Nothing to talk about on a NAS System - ext3 ist stable and there are issues with ext4 used in a RAID.
ext3 can be mounted on any OS in case of emergency with available drivers - those drivers still dont exist for ext4.
There is no reason for ext4 at the moment.

The ext4 issues are not connected to QNAP-devices. The trouble may be caused by the delayed allocation ext4 is using:
http://www.articlesnatch.com/Article/De ... nux/737485

The delayed allocation in Ext4 file system always poses some additional risks of data loss in situations where your system crashes before all the data has got written on the drive. The general situation where it might take place is an application replacing the contents of file without forcing write to disk with the fsync. Issues might occur if system gets crashed before actual write operation takes place. In such situations, the kernel of Ext4 file system would clear the contents of file before crash and thus the contents of file would completely lose. This behavior generally cause file system corruption and put you in need of Linux Recovery.

Add this to the fact, that most user are running RAID-Konfiguration with common Desktopdrives where the TLER/CCTL is set far to high and you have some unpredictble behavior resulting in lost data, or large files cut to 2GB like reported by some users in the german forum - returning to ext3 solved the troubles.

Greets Complicated

[schumaku]

There is no reason for ext4 at the moment.

Files >> 2 TB created e.g. by Time Machine is a very reasonable argument...

-Kurt.

[Complicated]

There is no reason for ext4 at the moment.

Files >> 2 TB created e.g. by Time Machine is a very reasonable argument...

Sure, but rather not that important for Homeusers. And I don't know if the argument outweights the RAID-Trouble for the Businessusers and how much Time Machine is in use there
We are talking about Filezize!

[Q]

delayed allocation:
according to various sources that was improved with kernel 2.6.30. as the new firmware 3.2.0 comes with kernel 2.6.30.6, it should be ok i think. but of course you're right, ext3 is more mature and if you want to be on the safe side it's the fs to choose.

driver:
i found a free software than can read ext4 within windows, check out viewtopic.php?f=50&t=22769.

[Complicated]

I found this interessting Quote of Linus Torvalds regarding ext4:

It literally does everything the wrong way around -- writing data later than the metadata that points to it. Whoever came up with that solution was a moron. No ifs, buts, or maybes about it. At least with ext3 it's not the default mode."

More about this issue and discussion:
http://www.linux-magazine.com/Online/Ne ... 3-and-Ext4

Linus Torvalds, Ted Ts'o, Alan Cox, Ingo Molnar, Andrew Morton and other Linux kernel developers are embroiled in a contentious discussion over the sense -- or nonsense -- of journaling and delayed allocation before a commit in the ext3 and ext4 filesystems. Heavy words are flying.

As it is stated there the ext3 FS writes the meta Data of the file and the file itself with a delay of 5 sek. - ext4 is delaying the writing of the physical Data up to 2 min.
For RAID-Systems this fact can lead to a complete different behavior then Single Disk Systems.
Add to this equation users (yes all of us ) and their creativity when it comes to file access scenario (iSCSI, video editing files stored on the NAS, Office Files stored on NAS (*.pst e.g.)).

So IMHO Qnap should consider to add a warning where the selection of the filesystem ist placed in the GUI or maybe where RAID ist placed. Most user would blame Qnap at this moment for something not being their fault when data corruption occurs in those scenarios.

Complicated

[lucas72]

@Complicated

That article is quite old (Mar 30, 2009) and refer mostly to kernel 2.6.29.

[tovalb]

will i notice any perfomanceboost on my ts-119 (single-drive), or should i stay with the ext3?

Tom