| View previous topic :: View next topic |
| Author |
Message |
nikulinpi
n00b
Joined: 12 Jan 2007
Posts: 38
|
 Posted: Fri Mar 25, 2022 2:43 pm Post subject: How stable is F2FS now?Have they got fsck to a usable level? |
 |
|
I am thinking of switching for f2fs for my new SSD, as it looks more, and more impressive in benchmarks over ext4.
I am deterred by suggestions of it being prone to corruption, and weak fsck. Is it still the case? |
|
| Back to top |
|
 |
Goverp
Advocate
Joined: 07 Mar 2007
Posts: 2202
|
| Posted: Fri Mar 25, 2022 5:51 pm Post subject: |
|
|
I've been using F2FS for my root partition on my desktop machine for several years, and my laptop since I replaces its HDD with an SDD just over a year ago. Smartctl tells me my laptop has been powered on for 1080 hours, and had 99 unsafe powerdowns. I've had no issues at all, and no files in lost+found - perhaps less than I'd have had with an unsafe powerdown of ext4. I've also used it for several years on Raspberry Pi SD cards, no problems in the last 5 years at a guess.
AFAIK it still won't do a boot-time fsck without an initramfs script to do the same - that said since I set my machines up to do it, I've not needed to find out! That's on the laptop and desktop machines. The Pis run Arch linux, and I've not noticed neither boot-time fsck issues, nor indeed boot-time fscks!
_________________
Greybeard |
|
| Back to top |
|
|
figueroa
Advocate
Joined: 14 Aug 2005
Posts: 3007
Location: Edge of marsh USA
|
| Posted: Sat Mar 26, 2022 4:06 am Post subject: |
|
|
Goverp, have you really had 99 unsafe powrerdowns? That seems extraordinary.
_________________
Andy Figueroa
hp pavilion hpe h8-1260t/2AB5; spinning rust x3
i7-2600 @ 3.40GHz; 16 gb; Radeon HD 7570
amd64/23.0/split-usr/desktop (stable), OpenRC, -systemd -pulseaudio -uefi |
|
| Back to top |
|
|
Goverp
Advocate
Joined: 07 Mar 2007
Posts: 2202
|
| Posted: Sat Mar 26, 2022 9:30 am Post subject: |
|
|
figueroa,
Good point. I unthinkingly quoted what smartctl says. It's a laptop. Sometimes the battery runs out. That said, 99 would be about 1 every 4 days, and I think I'd remember that. As Neddy Seagoon has often pointed out, raw data in smart output is usually suspect 'cos it can be either coded or often part of multi-use fields - 99 is 0x63, so perhaps the number is a more realistic 6 or 3....
_________________
Greybeard |
|
| Back to top |
|
|
logrusx
Advocate
Joined: 22 Feb 2018
Posts: 2686
|
| Posted: Sat Mar 26, 2022 9:55 am Post subject: |
|
|
From Wikipedia F2FS page:
| Quote: | | F2FS divides the whole volume into a number of segments, each of which is fixed at 2 MB. A section is composed of consecutive segments, and a zone consists of a set of sections. By default, section and zone sizes are set to the same size, but users can easily modify the size with mkfs. |
Since there's a layer of abstraction at the firmware level, that makes no sense. Physically, any block can appear anywhere and there's no way of knowing that. It's at firmware's discretion.
Furthermore:
| Quote: | | F2FS splits the entire volume into six areas, and all except the superblock area consist of multiple segments as described below. |
That also doesn't make much sense for the same reason.
| Quote: | | In order to avoid misalignment between file system and flash storage, F2FS aligns the start block address of the CP with the segment size. It also aligns the Main Area start block address with the zone size by reserving some segments in the SSA area. |
Same as above. For everything that has to do with physical location, the above applies.
The rest seems somehow ordinary. It supports TRIM/FITRIM, which is very important for SSD, but all other popular formats support it too.
I'd say that, performance could be the only reason to use F2FS, if the difference is significant enough. Otherwise this FS was developed in other times and in my opinion its main design goals don't reflect the current reality.
If your only reference points are benchmarks, I'd say don't bother with it, except if you just want to play with it or do it out of boredom or curiosity or something. Of course the best would be to produce some raw data that could actually help others decide.
It doesn't look actively developed as well. Last change is from 3 years ago.
Regards,
Georgi |
|
| Back to top |
|
|
Goverp
Advocate
Joined: 07 Mar 2007
Posts: 2202
|
| Posted: Sat Mar 26, 2022 11:36 am Post subject: |
|
|
| logrusx wrote: | ...
It doesn't look actively developed as well. Last change is from 3 years ago.
...i |
Cobblers. A random selection of items from Phoronix lists performance in kernel 5.15 and 5.17 and coming in 5.18 and 5.19, and compression improvements in 5.14.
_________________
Greybeard |
|
| Back to top |
|
|
logrusx
Advocate
Joined: 22 Feb 2018
Posts: 2686
|
| Posted: Sat Mar 26, 2022 11:46 am Post subject: |
|
|
| Goverp wrote: | | logrusx wrote: | ...
It doesn't look actively developed as well. Last change is from 3 years ago.
...i |
Cobblers. A random selection of items from Phoronix lists performance in kernel 5.15 and 5.17 and coming in 5.18 and 5.19, and compression improvements in 5.14. |
I don't know what cobblers means. I can't understand what the meaning of the rest is either.
I checked with the git repo. Last change was from July 2019. The rest is various merges of the newer kernel versions into the development branch. Nothing related to F2FS since then. It's also badly documented. I guess the option to read the code remains, but I don't need to do that.
UPDATE: the above underlined is not true. I've checked the wrong repo, so disregard that part.
Regards,
Georgi |
|
| Back to top |
|
|
Goverp
Advocate
Joined: 07 Mar 2007
Posts: 2202
|
| Posted: Sat Mar 26, 2022 12:10 pm Post subject: |
|
|
| logrusx wrote: | ...
I don't know what cobblers means. I can't understand what the meaning of the rest is either.
... |
"Cobblers" means false. The rest means that the phoronix.com website contains a frequent series of F2FS articles identifying kernel changes for improved performance and function. For example, the most recent, which references a kernel pull request. There have been similar articles for kernel 5.16 and so forth. That looks like development to me. Perhaps you are looking at an obsolete git repository.
If by "active development" you mean new functions, I think most filesystems rapidly meet a similar state - once they deliver the filesystem API, functions are added only infrequently. (F2FS gained compression and encryption support a few years back, about the same time as ext4, using the same fscrypt interface. I'm not sure either have added new function since then.)
_________________
Greybeard |
|
| Back to top |
|
|
logrusx
Advocate
Joined: 22 Feb 2018
Posts: 2686
|
| Posted: Sat Mar 26, 2022 12:16 pm Post subject: |
|
|
| Goverp wrote: |
If by "active development" you mean new functions |
See the update in my previous replay. |
|
| Back to top |
|
|
|
Kernel & Hardware
