Veritas 5.0.1 Installation Guide, HP-UX 11i v3, First Edition, November 2009

2. Upgrade from HP-UX 11i v1 to HP-UX 11i v2. This upgrade will install Base-VXVM 4.1 by
default.
3. Upgrade from HP-UX 11i v2 to HP-UX 11i v3. This upgrade will install Base-VXVM 4.1 by
default.
For more information on upgrading to HP-UX 11i v3, see the HP-UX 11i v3 Installation and
Update Guide at: http://www.docs.hp.com.
4. If patches to VxVM 5.0.1 on HP-UX 11i v3 are required, the patches must be applied prior
to upgrading the product.
5. Install the VxVM 5.0.1 package.
# swinstall –x autoreboot=true -s <depot_path> Base-VxVM-501
Upon successful completion of the upgrade, VxVM 4.1 will be replaced by VxVM 5.0.1. Disk
groups with older disk group versions (90 or 120) will be accessible after the upgrade.
I/O Robustness Recommendations
To take full advantage of I/O retry mechanisms in HP-UX and ensure the highest levels of
robustness in SAN environments, HP recommends the following steps:
• Disable VxVM Powerfail Timeout (PFTO) feature on all disk groups using either of the
following commands:
vxdisk -g <dg_name> set <disk_name> pftostate = {enabled | disabled}
vxpfto -g <dg_name> -o pftostate = {enabled | disabled}
This essentially allows HP-UX's I/O stack to perform sufficient retries on I/Os and not
prematurely timeout the I/O request. Any PFTO setting currently existing is ignored after
disabling the PFTO feature. You can use the commands mentioned above to re-enable PFTO,
if necessary. For more information, see pfto(7), vxpfto(1M), vxdisk(1M) man pages.
• Enable DMP time-based LUN retries using the following command:
vxdmpadm settune dmp_lun_retry_timeout=120
This system-wide tunable instructs DMP to retry failed I/O requests on a LUN for the time in
seconds specified. The vxdmpadm manpage has not been updated for this release. A brief
description of the new parameter, dmp_lun_retry_timeout is described as follows:
dmp_lun_retry_timeout
When all paths to a disk fail, there may be certain paths that have a temporary failure and are
likely to be restored soon. If I/Os are not retried for a period of time, the I/Os may be failed to
the application layer even though some paths are experiencing a transient failure. For more
robust handling of such transient errors, the dmp_lun_retry_timeout tunable provides a
time-based retry mechanism at the LUN level. If the tunable is set to a non-zero value, I/Os to a
disk with all failed paths will be retried until the dmp_lun_retry_timeout interval expires
or until the I/O succeeds on one of the paths, whichever happens first.
Use the following command to set and retrieve the tunable:
vxdmpadm gettune dmp_lun_retry_timeout
vxdmpadm settune dmp_lun_retry_timeout=<interval>
Where the value for <interval> is in seconds.
The default value of this tunable is 0, which means that the paths are probed only once. HP
recommends a value of 120 seconds.
You need to set the pftostate and dmp_lun_retry_timeout tunables only once as the
values you set are persistent across reboots.
If these tunables are not set as recommended, certain transient I/O error conditions could result
in interruptions to volume availability.
I/O Robustness Recommendations 65