How to temporarily stop the jbd2-demon of the ext4 file system?

JZL240I-U · 06-28-2017, 07:51 AM

I have a hard disk which is from time to time annoying me: it starts "ticking" like a quietly and steadily beating heart (i.e. it is moving ist magnetic heads about 88 times / minute). Searching for the cause is documented in this lengthy thread: http://www.linuxquestions.org/questi...rt-4175594770/

One possible culprit might be jbd2:

Code:

PC:/home/me # dmesg|tail
[ 4283.984763] jbd2/sda6-8(954): WRITE block 453324480 on sda6 (8 sectors)
[ 4283.985042] jbd2/sda6-8(954): WRITE block 453324488 on sda6 (8 sectors)
[ 4286.696612] kworker/u16:3(4218): WRITE block 3769248 on sdc2 (8 sectors)
[ 4289.512775] jbd2/sda6-8(954): WRITE block 453324496 on sda6 (8 sectors)
[ 4289.512802] jbd2/sda6-8(954): WRITE block 453324504 on sda6 (8 sectors)
[ 4289.513094] jbd2/sda6-8(954): WRITE block 453324512 on sda6 (8 sectors)
[ 4291.899308] bash(4579): READ block 49416608 on sdc2 (32 sectors)
[ 4291.899374] bash(4578): READ block 49542552 on sdc2 (32 sectors)
[ 4291.911403] dmesg(4578): READ block 49542584 on sdc2 (104 sectors)
[ 4291.911442] tail(4579): READ block 49416640 on sdc2 (104 sectors)

To make sure, I want to temporarily switch off jbd2 but I have no idea how to proceed. Google did not help *sigh*. Anybody here with this particular piece of knowledge?

MensaWater · 06-28-2017, 08:08 AM

jdb2 is run by the kernel for the Journaling of ext4 filesystems. Since it is started by the kernel I don't think it can be turned off.

However, this post suggests you can change the timing by doing a workaround:

Quote:

Adding commit=60 to the partition's entry in fstab increases the interval to 30-45 seconds.

If you change the timing and see if it affects how often you hear the clicks it might help prove this is your issue.

pan64 · 06-28-2017, 08:10 AM

probably you can find something useful here: https://bbs.archlinux.org/viewtopic.php?id=113516&p=2

JZL240I-U · 06-29-2017, 04:26 AM

I'll have to wait until the disk is doing it again. I'll report then. Thanks for your inputs.

JZL240I-U · 11-05-2017, 10:39 AM

Sorry, but no. pm-utils are not installed and /var/log/crond.log, /var/log/everything.log and /var/log/auth.log don't even exist, so that can't be it either. Funny. Any other ideas?

MensaWater · 11-06-2017, 08:41 AM

If your distro/version uses systemd it may not have syslog (/var/log/...) files but instead rely on journald. (On RHEL7 they use both.)

You can use "journalctl" to view logging done by journald. "man journalctl" will give you all the usage options.
You can run "journalctl -l" to see entire lines.

JZL240I-U · 11-06-2017, 12:34 PM

Hi MensaWater,

Yes, openSUSE uses journald, thanks.

But those are sh*tty man-pages (and none of your fault, I'd guess). Darn dot.abbreviations (.SH .SE .PP ...) with no legend, not to mention the hare-brained formatting -- grrr.

Problem is, it happened last time yesterday, so it's many thousands of lines back in the log. Do you have any idea what we are looking for so I can "grep" for the culprit?