IO timeout change in PV - how to change and can it be done online?

[sanwin]

Hi,
Could anyone can guide me in changing the IO timeout value for a PV.

EMC has recommended to change the IO timeout value to overcome our sap performance issues.

so i tried to find the procedure to do the change of value for IO timeout parameter.

Can it be done online?

What effect it will have if we change online?

Or how should we modify the value and what are the prerequisites to do that.


Waiting for your comments and help.



PV Name /dev/dsk/c13t1d7
PV Name /dev/dsk/c7t1d7 Alternate Link
VG Name /dev/vgIAD
PV Status available
Allocatable yes
VGDA 2
Cur LV 2
PE Size (Mbytes) 16
Total PE 6399
Free PE 0
Allocated PE 6399
Stale PE 0
IO Timeout (Seconds) default
Autoswitch On
Proactive Polling Off




Regards,
sanwin

[sujit kumar singh]

hi


yes this can be done online.
#pvchange -t 120 /dev/dsk/cXtYdZ

this sets the Time for IO timeout attempts from LVM in which if a response is not recd form the PV or the Path of the PV , that will be marked as failed ands if that is a pTH failure and Alternate Paths are configured then the Alternate path will be used for IO.

This can be done online.
regards
sujit

[Steven E. Protter]

Shalom,

lvchange
http://docs.hp.com/en/B9106-90008/lvchange.1M.html

pvchange
http://docs.hp.com/en/B2355-60130/pvchange.1M.html

As to how an EMC LUN will react, I'm not sure. I see no indication requiring downtime. You should be able to do this hot.

Be sure you are comfy with EMC recommended changes. I once saw a case where a simple change was made in a customer system and 18 months of diagnosis for performance issues came next.

SEP

[F Verschuren]

see man pvchange:

/usr/sbin/pvchange [-A autobackup] -t IO_timeout


-t IO_timeout Set IO_timeout for the physical volume to the
number of seconds indicated. An IO_timeout
value of zero (0) causes the system to use
the default value supplied by the device
driver associated with the physical device.
IO_timeout is used by the device driver to


determine how long to wait for disk
transactions to complete before concluding
that an IO request can not be completed (and
the device is offline or unavailable).

[Ganesan R]

Hi Sanwin,

Usually we leave the I/O timeout value to default which is derived from the device driver unless if it is needed to change for heavily loaded disk or recommanded by the vendor.

Use pvchange -t to change from default value.

There won't be any impact. For example, if you set to 60sec, system will wait till 60sec to declare the disk/path as offline if any I/O failed to that disk/path.

[Duncan Edmonstone]

>> EMC has recommended to change the IO timeout value to overcome our sap performance issues.


I'd be interested to know how EMC expect an IO timeout (which only comes into play when a PV cannot be accessed and is measured in multiples of seconds) is going to effect SAP performance...

Have you asked EMC to explain *why* they think changing this parameter will help?

HTH

Duncan

[Steven E. Protter]

Shalom again.

Let me echo Duncan's remark.

Changing this parameter will not improve performance.

Other EMC parms might help, would a thorough review of I/O layout and LUN configuration.

Heavy write disks/LUNS should be raid 1 or raid 10.

SEP