| Author |
Message |
dibl
|
|
|
Post subject: SOLVED /etc/fstab -- ext4 mount option "commit"
Posted: 02.04.2011, 21:51
|
|

Joined: 2010-09-12
Posts: 302
Location: Dayton, Ohio, USA
Status: Offline
|
|
I just noticed this -- I don't know how long this condition has been present. I use a "commit=300" mount option on my ext4 partitions, because (a) my desktop system runs on a UPS, and (b) two partitions are on a large SSD. Here's the mount line for the root filesystem in /etc/fstab:
Code:
UUID=c70374e7-97b3-4972-a704-285e9343f776 / ext4 defaults,noatime,errors=remount-ro,barrier=0,discard,commit=300 0 1
But, when I run the mount command to see the mounted partitions, it has apparently added to a "commit=0" which, according to the manual, reverts to the default 5-second commit cycle. Running "mount", here is what I see for the root filesystem:
Code:
/dev/sda1 on / type ext4 (rw,noatime,errors=remount-ro,barrier=0,discard,commit=300,commit=0)
What is going with ext4?
Code:
don@aptosidbox:~$ infobash -v3
Host/Kernel/OS "aptosidbox" running Linux 2.6.38-2.slh.2-aptosid-amd64 x86_64 [ aptosid 2010-03 Ἀπάτη - kde-lite - (201012262151) ]
CPU Info 8x Intel Core i7 950 @ 8192 KB cache flags( sse3 ht nx lm vmx ) clocked at [ 1600.000 MHz ]
Videocard nVidia GF100 [GeForce GTX 480] X.Org 1.9.5 [ 1920x1200@50.0hz ]
Network cards Marvell 88E8056 PCI-E Gigabit
Processes 229 | Uptime 4days | Memory 3092.1/5978.8MB | HDD OCZ-REVODRIVE,OCZ-REVODRIVE,WDC WD1002FAEX-0,WDC WD1002FAEX-0,KINGSTON SS100S2 Size 2136GB (2%used) | GLX Renderer GeForce GTX 480/PCI/SSE2 | GLX Version 4.1.0 NVIDIA 270.30 | Client Shell | Infobash v3.35
|
Last edited by dibl on 07.04.2011, 14:10; edited 2 times in total
|
| |
|
|
|
 |
slh
|
|
Post subject: RE: /etc/fstab -- ext4 mount option "commit" weird
Posted: 02.04.2011, 22:01
|
|

Joined: 2010-08-25
Posts: 585
Status: Offline
|
|
Please try a shorter commit intervall first, something like 10s for a start and observe what happens if you play with the values.
That said, personally I'd suggest to stay with 5 seconds or at least something in that ball park, 5 minutes is a whole long time to mess up your filesystem - not only on power outtages, but also if your kernel goes south - and modern SSDs aren't that fragile in regards to write cycles anymore. |
Last edited by slh on 02.04.2011, 22:09; edited 1 time in total
|
| |
|
|
|
 |
dibl
|
|
Post subject: RE: /etc/fstab -- ext4 mount option "commit" weird
Posted: 02.04.2011, 22:06
|
|

Joined: 2010-09-12
Posts: 302
Location: Dayton, Ohio, USA
Status: Offline
|
|
| Good suggestion -- I'll try it and report back. |
|
|
| |
|
|
|
 |
dibl
|
|
Post subject: RE: /etc/fstab -- ext4 mount option "commit" weird
Posted: 02.04.2011, 22:14
|
|

Joined: 2010-09-12
Posts: 302
Location: Dayton, Ohio, USA
Status: Offline
|
|
| Stranger yet. I edited /etc/fstab to change the commit on the root filesystem to 10, and issued
Code:
mount -a
The mount command shows that the commit=300,commit=0 has not changed. Now I will reboot the system. |
Last edited by dibl on 03.04.2011, 13:19; edited 1 time in total
|
| |
|
|
|
 |
dibl
|
|
Post subject: RE: /etc/fstab -- ext4 mount option "commit" weird
Posted: 02.04.2011, 22:24
|
|

Joined: 2010-09-12
Posts: 302
Location: Dayton, Ohio, USA
Status: Offline
|
|
So, after a reboot, it picked up the new commit interval. But ...
Code:
/dev/sda1 on / type ext4 (rw,noatime,errors=remount-ro,barrier=0,discard,commit=10,commit=0)
|
|
|
| |
|
|
|
 |
Lat
|
|
Post subject: RE: /etc/fstab -- ext4 mount option "commit" weird
Posted: 02.04.2011, 23:12
|
|

Joined: 2010-09-19
Posts: 147
Status: Offline
|
|
show us
Code:
grep sda1 /etc/fstab
|
|
|
| |
|
|
|
 |
dibl
|
|
Post subject: RE: /etc/fstab -- ext4 mount option "commit" weird
Posted: 02.04.2011, 23:28
|
|

Joined: 2010-09-12
Posts: 302
Location: Dayton, Ohio, USA
Status: Offline
|
|
That won't work because I mount by-uuid. However, here's the line, as I have it at the moment:
Code:
UUID=c70374e7-97b3-4972-a704-285e9343f776 / ext4 defaults,noatime,errors=remount-ro,barrier=0,discard 0 1
HOWEVER, after rebooting, here is what is mounted (output of the mount command):
Code:
/dev/sda1 on / type ext4 (rw,noatime,errors=remount-ro,barrier=0,discard,commit=0)
Notice that it has added "commit=0" to my mount line, even though I did not use it. |
|
|
| |
|
|
|
 |
dibl
|
|
Post subject:
Posted: 07.04.2011, 14:08
|
|

Joined: 2010-09-12
Posts: 302
Location: Dayton, Ohio, USA
Status: Offline
|
|
FIXED: The issue is triggered by a re-mount that happens during the boot process, and then the power management rule at /usr/lib/pm-utils/power.d/journal-commit is invoked. The default commit value for "AC" is commit=0, and that is appended to the existing mount. So, I edited the file and changed the value to commit=120 (same as I now have it set in /etc/fstab). Now the three ext4 filesystems on my system are mounted correctly at the end of the boot process.
FYI, this file is also invoked on resume from suspend, so the issue is seen in that situation also. |
|
|
| |
|
|
|
 |
|
|
|