You are not logged in.
I know that this is a delicate issue since it can kill a HDD, so I'm searching for some insight. I have laptop-mode-tools installed and running. When I'm on battery my HDD keeps parking and unparking heads, producing the characteristic "click" that is like a funeral bell for the drive... Through iotop I saw that, when in idle, the typical disk activity is disk write by some process (mainly plasma-desktop and skype); then "click", heads are parked, and after some seconds there's a new disk write activity which triggers an unparking, and so on resulting in a continuous park/unpark. As far as I understand, one of the features of laptop-mode is to set a time period during which data to be written on disk is kept in memory to prevent fequentdisk activity and the above described situation. I think taht I set it to 10 minutes when on battery by the LM_BATT_MAX_LOST_WORK_SECONDS setting in my /etc/laptop-mode/llaptop-mode-conf, but it actually doesn't seem to give many benefits. I'm really puzzled since I can't figure out how to solve this problem and save my HDD from premature death. Here's my laptop-mode.conf (purged from comments), in case someone has some good hint to give me:
ENABLE_LAPTOP_MODE_TOOLS=1
LOG_TO_SYSLOG=1
DEBUG=0
ENABLE_LAPTOP_MODE_ON_BATTERY=1
ENABLE_LAPTOP_MODE_ON_AC=1
ENABLE_LAPTOP_MODE_WHEN_LID_CLOSED=0
ENABLE_AUTO_MODULES=1
MINIMUM_BATTERY_CHARGE_PERCENT=3
DISABLE_LAPTOP_MODE_ON_CRITICAL_BATTERY_LEVEL=1
HD="/dev/[hs]d[abcdefgh]"
PARTITIONS="auto /dev/mapper/*"
ASSUME_SCSI_IS_SATA=1
LM_BATT_MAX_LOST_WORK_SECONDS=600
LM_AC_MAX_LOST_WORK_SECONDS=360
CONTROL_READAHEAD=1
LM_READAHEAD=3072
NOLM_READAHEAD=128
CONTROL_NOATIME=0
USE_RELATIME=1
CONTROL_HD_IDLE_TIMEOUT=1
LM_AC_HD_IDLE_TIMEOUT_SECONDS=7200
LM_BATT_HD_IDLE_TIMEOUT_SECONDS=20
NOLM_HD_IDLE_TIMEOUT_SECONDS=7200
CONTROL_HD_POWERMGMT=1
BATT_HD_POWERMGMT=127
LM_AC_HD_POWERMGMT=254
NOLM_AC_HD_POWERMGMT=254
CONTROL_HD_WRITECACHE=0
NOLM_AC_HD_WRITECACHE=1
NOLM_BATT_HD_WRITECACHE=0
LM_HD_WRITECACHE=0
CONTROL_MOUNT_OPTIONS=1
LM_DIRTY_RATIO=60
NOLM_DIRTY_RATIO=40
LM_DIRTY_BACKGROUND_RATIO=1
NOLM_DIRTY_BACKGROUND_RATIO=10
DEF_UPDATE=5
DEF_XFS_AGE_BUFFER=15
DEF_XFS_SYNC_INTERVAL=30
DEF_XFS_BUFD_INTERVAL=1
DEF_MAX_AGE=30
XFS_HZ=100
LM_SECONDS_BEFORE_SYNC=15Thanks.
Last edited by snack (2012-11-10 16:46:02)
Offline
Try setting BATT_HD_POWERMGMT to 254 or increasing LM_BATT_HD_IDLE_TIMEOUT_SECONDS.
Last edited by lucke (2012-11-08 23:57:25)
Offline
I want BATT_HD_POWERMGMT to stay on 127 since it is the highest value that permits spindown. Also, increasing LM_BATT_HD_IDLE_TIMEOUT_SECONDS should (if I'm correct) increase the idle time after which the disk spins down; so it is not related to head parking frequency. In fact, I don't want the heads to be parked less frequently, since it is the correct behavior to save energy; I'd like them to be unparked less frequently, since that consumes energy and triggers a subsequent parking.
BTW, my HDD type is:
$ sudo hdparm -I /dev/sda
/dev/sda:
ATA device, with non-removable media
Model Number: TOSHIBA MK3252GSX
Serial Number: 68E6P0RGT
Firmware Revision: LV010A
Transport: Serial, ATA8-AST, SATA 1.0a, SATA II Extensions, SATA Rev 2.5, SATA Rev 2.6
Standards:
Supported: 8 7 6 5
Likely used: 8
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 268435455
LBA48 user addressable sectors: 625142448
Logical Sector size: 512 bytes
Physical Sector size: 512 bytes
device size with M = 1024*1024: 305245 MBytes
device size with M = 1000*1000: 320072 MBytes (320 GB)
cache/buffer size = 8192 KBytes
Capabilities:
LBA, IORDY(can be disabled)
Queue depth: 32
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Advanced power management level: 254
DMA: sdma0 sdma1 sdma2 mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
* Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* NOP cmd
* DOWNLOAD_MICROCODE
* Advanced Power Management feature set
SET_MAX security extension
* 48-bit Address feature set
* Device Configuration Overlay feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
* SMART error logging
* SMART self-test
* General Purpose Logging feature set
* WRITE_{DMA|MULTIPLE}_FUA_EXT
* 64-bit World wide name
* IDLE_IMMEDIATE with UNLOAD
* Gen1 signaling speed (1.5Gb/s)
* Native Command Queueing (NCQ)
* Host-initiated interface power management
* Phy event counters
* Idle-Unload when NCQ is active
* DMA Setup Auto-Activate optimization
Device-initiated interface power management
* Software settings preservation
* SMART Command Transport (SCT) feature set
* SCT Error Recovery Control (AC3)
* SCT Features Control (AC4)
* SCT Data Tables (AC5)
Security:
Master password revision code = 65534
supported
not enabled
not locked
frozen
not expired: security count
not supported: enhanced erase
128min for SECURITY ERASE UNIT.
Logical Unit WWN Device Identifier: 5000039112403511
NAA : 5
IEEE OUI : 000039
Unique ID : 112403511
Checksum: correctOffline
CONTROL_HD_POWERMGMT=0
BATT_HD_POWERMGMT=255
sudo hdparm -B255 /dev/sda
yaourt -S systemd-hddapm
sudo systemctl enable hddapm
Offline
I suppose that you are suggesting me to disable laptop-mode's HDD management and install the corresponding systemd tool... Yes, I could give it a try but there's not so much documentation on how it works and in the end I like laptop-mode and I would like to understand why it is not working properly and maybe fix it, not just throw it away.
Offline
You can use "dstat -d --top-bio" to see when your disk is being read from or written to and what process is doing it. Chances are your disk is waking up because something is reading from it - laptop_mode can only batch writes.
-edit-
If those are writes indeed that are waking the disk, make sure /proc/sys/vm/laptop_mode is indeed non-zero and /proc/sys/vm/dirty_writeback_centisecs and /proc/sys/vm/dirty_expire_centisecs are set to 60000 (10 min).
Last edited by lucke (2012-11-09 16:37:41)
Offline
As I wrote on my first post, I already did it using iotop. I found that most disk activity comes from skype and kdeinit plasma-desktop, and since it is mostly write activity I would expect that write caching configured by laptop-mode would prevent the actual write on disk and the related load/unload cycles...
Offline
Please note my edit.
I've just tested laptop-mode on my desktop pc and I do get writes fairly often. Last time I was monitoring it on my netbook, it was working okay. Maybe it's some kernel regression? You could try on linux-lts and/or some old kernel from Arch's Rollback Machine. Or maybe reading http://www.mjmwired.net/kernel/Document … p-mode.txt would give you some idea.
Offline
Thanks lucke, you've been kind to add a post since edits are not notified and I was missing your edit. I'll try your suggestions this evening as I get back home.
Offline
I tested what lucke suggested and here's the result:
$ cat /proc/sys/vm/laptop_mode
5
$ cat /proc/sys/vm/dirty_writeback_centisecs
60000
$ cat /proc/sys/vm/dirty_expire_centisecs
60000so I think that settings are correct. I verified that the real culprit is "kdeinit4: plasma-desktop [kdeinit]", which always trigger the disk write, while skype's writes seems to be correctly cached. I'm not an expert but I think that maybe plasma-desktop has too much disk flush instructions, or something like that...
About the kernel: I'm on linux-ck with bfq scheduler. I tried also the cfq scheduler and the stock linux-lts, but the behavior is always the same.
I'm going to install and try lxde to nail down the problem to KDE.
Offline
OK, I confirm that it is a KDE issue. On lxde there's no such anomalous behavior. I don't know however how to dig further, maybe the best thing is to open a bug report on KDE bug tracker. Any other idea?
Offline
laptop_mode works indeed on my netbook. Firefox likes to write stuff often, though, as it was doing on the desktop (I was using Opera last time I checked how it works on my netbook). I'm a KDE user and I don't get writes from plasma-desktop, so it's probably one of your widgets (I use panels with a pretty regular set of widgets). First step would be to find out which one.
Offline
Your suggestion was a great shot, lucke: the trouble comes from the complex-plotter palsmoid. Removing it everything works normally. Now I have to figure out why it does so much disk activity and eventually report a bug upstream (if upstream still exists...)
Last edited by snack (2012-11-10 16:45:27)
Offline