Multipathing allows you to configure multiple network paths between the host and storage system. If one path fails, traffic continues on the remaining paths.

Non-ASA configurations

For non-ASA configurations, there should be two groups of paths with different priorities. The paths with higher priorities are Active/Optimized. This means they are serviced by the controller where the aggregate is located. The paths with lower priorities are active but non-optimized because they are served from a different controller. The non-optimized paths are only used when optimized paths are not available.

Example

The following example displays the correct output for an ONTAP LUN with two Active/Optimized paths and two Active/Non-Optimized paths.

The path priorities are displayed against the Access State section for each LUN in the OS native mpathadm show lu <LUN> command.

All SAN Array configurations

All SAN Array (ASA) configurations optimize all paths to a given LUN, keeping them active. This improves performance by serving I/O operations through all paths at the same time.

Example

The following example displays the correct output for an ONTAP LUN.

The path priorities are displayed against the Access State section for each LUN in the OS native mpathadm show lu <LUN> command.

All SAN Arrays (ASA) configurations are supported beginning in ONTAP 9.8 for Solaris Hosts.

Recommended settings

Following are some parameter settings that are recommended for Solaris 11.3 SPARC and x86_64 with ONTAP LUNs. These parameter values are set by Host Utilities.

Parameter	Value
throttle_max	8
not_ready_retries	300
busy_retries	30
reset_retries	30
throttle_min	2
timeout_retries	10
physical_block_size	4096

Recommended settings for MetroCluster

By default, the Solaris operating system will fail I/Os after 20 seconds if all paths to a LUN are lost. This is controlled by the fcp_offline_delay parameter. The default value for fcp_offline_delay is appropriate for standard ONTAP clusters. However, in MetroCluster configurations, the value of fcp_offline_delay must be increased to 120s to ensure that I/O does not prematurely time out during operations including unplanned failovers.

Oracle Solaris virtualization

Solaris virtualization options include Solaris Logical Domains (also called LDOMs or Oracle VM Server for SPARC), Solaris Dynamic Domains, Solaris Zones, and Solaris Containers. These technologies have been rebranded generally as "Oracle Virtual Machines" despite the fact that they are based on very different architectures.
In some cases, multiple options can be used together such as a Solaris Container within a particular Solaris Logical Domain.
NetApp generally supports the use of these virtualization technologies where the overall configuration is supported by Oracle and any partition with direct access to LUNs is a supported configuration. This includes root containers, LDOM I/O domains, and LDOM’s using NPIV to access LUNs.
Partitions and/or virtual machines which use only virtualized storage resources, such as a vdsk, do not need specific qualification as they do not have direct access to NetApp LUNs.

Recommended settings for virtualization

When LUNs are used as virtual disk devices within an LDOM, the source of the LUN is masked by virtualization and the LDOM will not properly detect the block sizes. To prevent this issue, the LDOM operating system must be patched for Oracle Bug 15824910 and a vdc.conf file must be created that sets the block size of the virtual disk to 4096. See Oracle Doc 2157669.1 for more information.

To verify the patch, do the following:

Steps

Create a zpool.
Run zdb -C against the zpool and verify that the value of ashift is 12.

If the value of ashift is not 12, verify that the correct patch was installed and recheck the contents of vdc.conf.

Do not proceed until ashift shows a value of 12.

Patches are available for Oracle bug 15824910 on various versions of Solaris. Contact Oracle if assistance is required in determining the best kernel patch.

Recommended settings for SnapMirror active sync

In order to verify that the Solaris client applications are non-disruptive when an unplanned site failover switchover occurs in a SnapMirror active sync environment, you must configure the following setting on the Solaris 11.3 host. This setting overrides the failover module f_tpgs to prevent the execution of the code path that detects the contradiction.

Beginning with ONTAP 9.9.1, SnapMirror active sync setting configurations are supported in the Solaris 11.3 host.

Follow the instructions to configure the override parameter:

Steps

Create the configuration file /etc/driver/drv/scsi_vhci.conf with an entry similar to the following for the NetApp storage type connected to the host:
```
scsi-vhci-failover-override =
"NETAPP  LUN","f_tpgs"
```
Use the devprop and mdb commands to verify that the override parameter has been successfully applied:

root@host-A:~# devprop -v -n /scsi_vhci scsi-vhci-failover-override scsi-vhci-failover-override=NETAPP LUN + f_tpgs root@host-A:~# echo "*scsi_vhci_dip::print -x struct dev_info devi_child | ::list struct dev_info devi_sibling| ::print struct dev_info devi_mdi_client| ::print mdi_client_t ct_vprivate| ::print struct scsi_vhci_lun svl_lun_wwn svl_fops_name"| mdb -k
```
svl_lun_wwn = 0xa002a1c8960 "600a098038313477543f524539787938"
svl_fops_name = 0xa00298d69e0 "conf f_tpgs"
```

SAN hosts and cloud clients manuals ( CA08872-021 )

Use Solaris 11.3 with ONTAP

SAN booting

Multipathing