VSCSI_HOST Error on Virtual I/O Server

Resolving The Problem

Below is a list of the most frequently seen Detail Data and recommendation based on module and rc value.

1. module: add_child rc: 0000000000000016

2. module: target_report_luns rc: 00000000FFFFFFD8

1.

module: add_child rc: 0000000000000016

LABEL:  VSCSI_HOST
IDENTIFIER: DF63A4FE

...snip...

Detail Data
ADDITIONAL INFORMATION
module: add_child   rc: 0000000000000016   location: 00001318
data:  E40000C

Return code 16 is commonly due to disk I/O problem.

Virtual SCSI requires read/write access to all backing devices (i.e. hdisk, logical volume, etc) exported thru the vhost.
The error may be logged when one or more existing backing devices are inaccessible by the VIOS or when trying to create a new Virtual Target Device (VTD) using a backing device that is inaccessible by the VIOS.

RECOMMENDATION

Ensure all existing VTDs exported thru the vhost are in Available state.

To list VTD names in a given vhost, run:

$ lsmap -vadapter vhost0|more

SVSA            Physloc                              Client Partition ID
--------------- ------------------------------------ ------------------
vhost0          U9117.MMA.101F1A0-V1-C20             0x00000007

VTD                   client_rtvg
Status                Available            <--------
LUN                   0x8100000000000000
Backing device        hdisk7
Physloc
Mirrored              N/A
...

For any VTD that may show "Status Defined", determine the corresponding backing device (hdisk7 in this example) and ensure it is accessible by the VIOS. For example, this can be done by running:

$ oem_setup_env
# lquerypv -h /dev/hdisk7 80 10  
00000080   00C1F1A0 E5EC0D5A 00000000 00000000  |.......Z........|

or

# dd if=/dev/hdisk7 of=/tmp/iotest count=1000


1000+0 records in.
1000+0 records out.

If either command hangs, returns back to the # prompt, or fails with some kind of error, that indicates the device is not accessible by the VIOS. That issue must be corrected first.

For VIOS storage connectivity problems, please, contact your storage support.

Once the commands return output similar to the above, try changing the VTD status back to "Available", by running

$ cfgdev -dev <VTD_name>

If the VSCSI_HOST error is logged when attempting to create a new VTD (i.e. mkvdev -vdev hdisk# -vadapter vhost#), verify the backing device specified in -vdev option is accessible by the VIOS before re-attempting mkvdev operation again.

2.

module: target_report_luns rc: 00000000FFFFFFD8

LABEL: VSCSI_HOST
IDENTIFIER:      DF63A4FE
...
Resource Name:   vhost18
...
Detail Data
ADDITIONAL INFORMATION
 module: target_report_luns rc: 00000000FFFFFFD8  location: 00000001

  data:  2 9100000000000000

This error suggests issue with the disk max_transfer size exported thru the vhost. For more details and recommendation see technote, Unable to See New VSCSI Disk on Existing vhost.