disk crunching at regular intervals

From: Vanilla Riddle (mail_at_invalid.com)
Date: 01/31/05


Date: Mon, 31 Jan 2005 07:25:50 +0000 (UTC)

EHLO. :-)

Long story short - I installed Debian Sarge on my compaq nx9020 laptop.
Right now it is running 2.6.10, no additional patches. Got it working
quite nicely except for one annoying detail - the disk keeps making
crunching sounds, which last about 2 seconds at somewhat regular
(5-10sec) intervals. This happens when the computer is doing absolutely
nothing. Sometimes it becomes a little bit more peaceful under the
console (my common working enviroment), but not enough to maintain the
drive in a hdparm-induced sleep-state for longer than 30 seconds. I'd be
very grateful if someone could help me find out what's causing this
mysterious activity. I have no prior experience investingating such
problems so any suggestions will be appreciated. Here's some information
that could be relevant:

The system is experiencing low system load, no significant running
services, except for a local postfix. Almost always working in the
console. Tried stopping all services including syslogd and klogd -
didn't help, activity persists.

I have only one partition, mounted as '/', ext3 filesystem.

some disk info:
ICH4: IDE controller at PCI slot 0000:00:1f.1
ICH4: chipset revision 3
ICH4: not 100% native mode: will probe irqs later
    ide0: BM-DMA at 0x1810-0x1817, BIOS settings: hda:DMA, hdb:pio
    
hda: ST94019A, ATA DISK drive
hda: max request size: 1024KiB
hda: 78140160 sectors (40007 MB) w/2048KiB Cache, CHS=16383/255/63,
UDMA(100)
hda: cache flushes supported <- what is this??

256MB DDR RAM, 32MB stolen video memory, swap as a static 512MB swapfile
at /var/swap/0. Thought this could be causing it and that I'd be
probably better off with a swap partition, but swapoff -a didn't stop
the crunching sounds.

I tried to find a program that would show what process is currently
writing to disk, found only one called 'atop', unfortunately it requires
kernel patching and the newest supported version is 2.6.8. I booted it
and I think I have spotted the culprit - 'kjournald' and 'pdflush' seem
to be causing this. But both being the kernel's processes, I can't
figure out how to make them stop this. I don't really even know what
pdflush does.

I suppose this could have something to do with the message:
kjournald starting. Commit interval 5 seconds
that I get during bootup. But why should the journal be updated every
5secs when nothing is being written to the disk?

I also suspect (but these are just guesses) the following capabilities
of the kernel I compiled in:
1) "preemptible" - could this have something to do with it?
2) io scheduler: anticipatory - likewise?

Included below is the output of iostat with an update interval of a
second - i was doing nothing at that time, just sitting back and watching:

Linux 2.6.10 (shadowlands) 2005-01-31

Time: 08:23:45
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 3,37 89,67 111,43 260352 323536
Time: 08:23:46
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 3,00 0,00 24,00 0 24
Time: 08:23:47
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0
Time: 08:23:48
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0
Time: 08:23:49
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0
Time: 08:23:50
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 8,00 0 8
Time: 08:23:51
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 5,00 8,00 152,00 8 152
Time: 08:23:52
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0
Time: 08:23:53
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0
Time: 08:23:54
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0
Time: 08:23:55
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 8,08 0 8
Time: 08:23:56
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 2,97 0,00 23,76 0 24
Time: 08:23:57
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn
hda 0,00 0,00 0,00 0 0

This kind of shows the regularity we're dealing with.

So far, this is it. I've no idea what to try next. Apart from this the
system appears to be running fine, but the crunching sounds are very
annoying and I fear that this behaviour could shorten the drive's
lifespan. I repeat my plea for help, and thanks for reading.

democrux



Relevant Pages

  • Re: replacing failing system disk
    ... newfs all new partititions, mkswap on swap partition. ... copying the contents off fo failing disk will only aggravate it ... chroot to the / of hdb, ... with hda vs hdb vs both present etc..etc.. ...
    (Debian-User)
  • Re: need help booting from SATA in 2.4.32
    ... will not recognize this disk. ... SCSI device sda: 78140160 512-byte hdwr sectors ... Could you retry with it statically linked in the kernel? ... hda: attached ide-disk driver. ...
    (Linux-Kernel)
  • Re: How to Move from 1 drive to another? [Done!]
    ... old disk is hda; mbr is installed hda; bios allows ... mount it; and only now mkdir lower level directory ... - edit etc/lilo.conf - modify appropriately but install hdb ...
    (Debian-User)
  • hda: DMA timeout error
    ... I had a working raid-1 setupwith two western digital drives being ... sometimes hda sometimes hde) to plop out: ... hda: dma timeout error: status=0x58 {DriveReady SeekComplete ... Disk failure on hda, ...
    (Linux-Kernel)
  • FileSystemWatcher problem - multiple hits
    ... I am serializing an object to XML and writing it to disk. ... is causing me to process the change twice. ... notifications? ...
    (microsoft.public.dotnet.languages.csharp)