[SOLVED] Write udev rule to change harddisk timeout

EricTRA · 05-26-2010, 04:12 AM

Hi all,

I want (have to) change the timeout on the harddisks of our VMWare virtual machines in Debian 5. Right now I run a small script to set the timeout manually at boot using rc.local. I would like to have it applied using a udev rule but haven't got a clue where to start. I've been reading and searching but cannot get it to work.

Our vendor provided a script for Suse and RedHat. I've tried changing the RedHat script for Debian but it doesn't work at all.

Here's the udevinfo output of the disk in question.

Code:

 looking at device '/block/sda':
    KERNEL=="sda"
    SUBSYSTEM=="block"
    DRIVER==""
    ATTR{range}=="16"
    ATTR{removable}=="0"
    ATTR{size}=="16777216"
    ATTR{capability}=="12"
    ATTR{stat}=="    1206      829    37844     1664     1192     2335    28216      404        0     1724     2068"

  looking at parent device '/devices/pci0000:00/0000:00:10.0/host0/target0:0:0/0:0:0:0':
    KERNELS=="0:0:0:0"
    SUBSYSTEMS=="scsi"
    DRIVERS=="sd"
    ATTRS{device_blocked}=="0"
    ATTRS{type}=="0"
    ATTRS{scsi_level}=="3"
    ATTRS{vendor}=="VMware  "
    ATTRS{model}=="Virtual disk    "
    ATTRS{rev}=="1.0 "
    ATTRS{state}=="running"
    ATTRS{timeout}=="30"
    ATTRS{iocounterbits}=="32"
    ATTRS{iorequest_cnt}=="0x979"
    ATTRS{iodone_cnt}=="0x979"
    ATTRS{ioerr_cnt}=="0xa"
    ATTRS{modalias}=="scsi:t-0x00"
    ATTRS{evt_media_change}=="0"
    ATTRS{queue_depth}=="32"
    ATTRS{queue_type}=="simple"

  looking at parent device '/devices/pci0000:00/0000:00:10.0/host0/target0:0:0':
    KERNELS=="target0:0:0"
    SUBSYSTEMS==""
    DRIVERS==""

  looking at parent device '/devices/pci0000:00/0000:00:10.0/host0':
    KERNELS=="host0"
    SUBSYSTEMS==""
    DRIVERS==""

  looking at parent device '/devices/pci0000:00/0000:00:10.0':
    KERNELS=="0000:00:10.0"
    SUBSYSTEMS=="pci"
    DRIVERS=="mptspi"
    ATTRS{vendor}=="0x1000"
    ATTRS{device}=="0x0030"
    ATTRS{subsystem_vendor}=="0x0000"
    ATTRS{subsystem_device}=="0x0000"
    ATTRS{class}=="0x010000"
    ATTRS{irq}=="17"
    ATTRS{local_cpus}=="ff"
    ATTRS{local_cpulist}=="0-7"
ATTRS{modalias}=="pci:v00001000d00000030sv00000000sd00000000bc01sc00i00"
    ATTRS{enable}=="1"
    ATTRS{broken_parity_status}=="0"
    ATTRS{msi_bus}==""

  looking at parent device '/devices/pci0000:00':
    KERNELS=="pci0000:00"
    SUBSYSTEMS==""
    DRIVERS==""

I also don't know if I should add it to 50-udev.rules, 60-persistent-storage.rules or create a new rule file.

All help is greatly appreciated.

Kind regards,

Eric

EricTRA · 05-26-2010, 06:07 AM

Hi,

After some more reading and searching I put this construct in /etc/udev/rules.d/81-vmware-timeout.rules

Code:

ACTION=="add", KERNEL=="sda", SUBSYSTEM=="block", RUN+="/bin/sh -c 'echo 180 > /sys/$devpath/timeout'"

but it's not doing a thing. Can someone point me to the error or reason why it's not working?

Kind regards,

Eric

catkin · 05-26-2010, 06:32 AM

Three parts to this; is the match working, is the RUN command being run and if so is it doing what was intended?

You could write a shellscript and call that instead of "/bin/sh -c 'echo 180 > /sys/$devpath/timeout'". The shellscript could write its command line and the echo command to log.

EricTRA · 05-26-2010, 07:28 AM

Hi catkin,

Thanks for your reply. The udev rule isn't doing anything at all. The timeout value doesn't get changed. I currently do it using a script but would prefer putting it in a udev rule. Any ideas?

Kind regard,

Eric

mxwlpxwl · 06-10-2010, 01:16 PM

EricTra,

You are on the right path with the following

Code:

ACTION=="add", KERNEL=="sda", SUBSYSTEM=="block", RUN+="/bin/sh -c 'echo 180 > /sys/$devpath/timeout'"

testing the rule first with "/usr/bin/udevtest /block/sda" will allow you to see where you messed up. I'm no expert on debian, but you may want to try

Code:

ACTION=="add", KERNEL=="sda", SUBSYSTEM=="block", PROGRAM="/bin/sh -c 'echo 180 > /sys/$devpath/timeout'"

Bueno Suerte

EricTRA · 06-11-2010, 12:17 AM

Hello and Welcome to LinuxQuestions,

Thank you very much for your reply, I'll try it out and let you know. Have fun using Linux.

Kind regards,

Eric

EricTRA · 06-11-2010, 01:12 AM

Quote:

Originally Posted by mxwlpxwl

Code:

ACTION=="add", KERNEL=="sda", SUBSYSTEM=="block", PROGRAM="/bin/sh -c 'echo 180 > /sys/$devpath/timeout'"

Bueno Suerte

Hi,

No joy

Tried different rule files without any result, not restarting udev nor rebooting. It just doesn't get applied. Could this have to do with the machines being virtual?

[EDIT]
Just tried it on a physical server with Debian and doesn't work either.
[/EDIT]

Kind regards,

Eric

colucix · 06-11-2010, 04:38 AM

Hi Eric,

pay attention to the fact that $devpath is substituted by "/block/sda", so that the path of the timeout file should be "/sys$devpath/device/timeout" (your rule lacks the device directory). As previously suggested you can test the rule with udevtest (when the disk is mounted) and if the rule's action is:

Code:

... RUN+="/bin/sh -c 'echo 180 > /sys$devpath/device/timeout'"

you should see this at some point:

Code:

$ udevtest /block/sda
<omitted>
main: run: '/bin/sh -c 'echo 180 > /sys/block/sda/device/timeout''

<omitted>[/CODE]
You may also want to refine your rule to match against that particular device, since "KERNEL=="sda", SUBSYSTEM=="block", is a bit too generic. For example you can try to add the "vendor" attribute which identifies a VMware disk and/or the "size" attribute, if fixed.

Cheers!
Alex

EricTRA · 06-11-2010, 05:29 AM

Hola Alex,

It's working!!! Thank you so much. The basic stuff is what gets forgotten first I think. I didn't think about to what the devpath variable was pointing, feeling pretty green right now

Anyway, udevtest doesn't work any more since version 0.122 and has been replaced by udevadm test <device>, so

Code:

udevadm test /block/sda

returned this:

Code:

srvtest:/etc/udev/rules.d# udevadm test /block/sda
This program is for debugging only, it does not run any program,
specified by a RUN key. It may show incorrect results, because
some values may be different, or not available at a simulation run.

add_matching_files: unable to open '/lib/udev/rules.d': No such file or directory
parse_file: reading '/etc/udev/rules.d/50-udev.rules' as rules file
parse_file: reading '/etc/udev/rules.d/60-persistent-input.rules' as rules file
parse_file: reading '/etc/udev/rules.d/60-persistent-storage-tape.rules' as rules file
parse_file: reading '/etc/udev/rules.d/60-persistent-storage.rules' as rules file
parse_file: reading '/etc/udev/rules.d/60-persistent-v4l.rules' as rules file
parse_file: reading '/dev/.udev/rules.d/61-dev-root-link.rules' as rules file
parse_file: reading '/etc/udev/rules.d/70-persistent-cd.rules' as rules file
parse_file: reading '/etc/udev/rules.d/75-cd-aliases-generator.rules' as rules file
parse_file: reading '/etc/udev/rules.d/75-persistent-net-generator.rules' as rules file
parse_file: reading '/etc/udev/rules.d/80-drivers.rules' as rules file
parse_file: reading '/etc/udev/rules.d/81-vmware-timeout.rules' as rules file
parse_file: reading '/etc/udev/rules.d/91-permissions.rules' as rules file
parse_file: reading '/etc/udev/rules.d/95-late.rules' as rules file
import_uevent_var: import into environment: 'MAJOR=8'
import_uevent_var: import into environment: 'MINOR=0'
import_uevent_var: import into environment: 'DEVTYPE=disk'
import_uevent_var: import into environment: 'PHYSDEVPATH=/devices/pci0000:00/0000:00:10.0/host0/target0:0:0/0:0:0:0'
import_uevent_var: import into environment: 'PHYSDEVBUS=scsi'
import_uevent_var: import into environment: 'PHYSDEVDRIVER=sd'
udevtest: looking at device '/block/sda' from subsystem 'block'
udev_rules_get_name: add symlink 'block/8:0'
match_rule: '/sys/block/sda/whole_disk' does not exist
udev_node_mknod: mknod(/dev/.tmp-8-0, 060600, (8,0))
udev_node_mknod: chmod(/dev/.tmp-8-0, 060600)
udev_node_mknod: chown(/dev/.tmp-8-0, 0, 0)
run_program: 'scsi_id --export --whitelisted --device=/dev/.tmp-8-0'
run_program: '/lib/udev/scsi_id' returned with status 1
run_program: 'edd_id --export /dev/.tmp-8-0'
run_program: '/lib/udev/edd_id' (stderr) 'no kernel EDD support'
run_program: '/lib/udev/edd_id' returned with status 2
run_program: 'path_id /block/sda'
run_program: '/lib/udev/path_id' (stdout) 'ID_PATH=pci-0000:00:10.0-scsi-0:0:0:0'
run_program: '/lib/udev/path_id' returned with status 0
udev_rules_get_name: add symlink 'disk/by-path/pci-0000:00:10.0-scsi-0:0:0:0'
run_program: 'vol_id --export /dev/.tmp-8-0'
run_program: '/lib/udev/vol_id' (stderr) '/dev/.tmp-8-0: unknown volume type'
run_program: '/lib/udev/vol_id' returned with status 4
udev_rules_get_name: no node name set, will use kernel name 'sda'
udev_device_event: device '/block/sda' already in database, cleanup
udev_node_add: creating device node '/dev/sda', major=8, minor=0, mode=0660, uid=0, gid=6
udev_node_update_symlinks: update symlink 'block/8:0' of '/block/sda'
udev_db_get_devices_by_name: found index directory '/dev/.udev/names/block\x2f8:0'
update_link: found 1 devices with name 'block/8:0'
update_link: found '/block/sda' for 'block/8:0'
update_link: compare (our own) priority of '/block/sda' 0 >= 0
update_link: 'block/8:0' with target 'sda' has the highest priority 0, create it
udev_node_update_symlinks: update symlink 'disk/by-path/pci-0000:00:10.0-scsi-0:0:0:0' of '/block/sda'
udev_db_get_devices_by_name: found index directory '/dev/.udev/names/disk\x2fby-path\x2fpci-0000:00:10.0-scsi-0:0:0:0'
update_link: found 1 devices with name 'disk/by-path/pci-0000:00:10.0-scsi-0:0:0:0'
update_link: found '/block/sda' for 'disk/by-path/pci-0000:00:10.0-scsi-0:0:0:0'
update_link: compare (our own) priority of '/block/sda' 0 >= 0
update_link: 'disk/by-path/pci-0000:00:10.0-scsi-0:0:0:0' with target 'sda' has the highest priority 0, create it
udevtest: run: '/bin/sh -c 'echo 180 > /sys/block/sda/device/timeout''
udevtest: run: 'socket:@/org/kernel/udev/monitor'
srvtest:/etc/udev/rules.d#

which looked pretty perfect! After a reboot of my testserver the correct value was presented. Thanks again for your help. Moltes graciès!

Kind regards,

Eric

colucix · 06-11-2010, 06:55 AM

Hola! You're welcome! Glad to see it works now. And thank you for the notice about udevadm... I always forget it (old habits are hard to die).