does msa 2060 fail over if one iscsi path fail?

[idrissi]

does msa 2060 fail over if one iscsi path fails (for example one cable disconnected)? msa (active/active mode) is connected to 4 hosts using iscsi, i searched hpe documentation but didn't found any clear response, if you ahve any reference to hpe documentation please mention it.

[support_s]

System recommended content:

1. Advisory: HPE MSA 2060 Storage - FIRMWARE UPDATE REQUIRED to Prevent Drive Failures on ST12000NM008G HDDs

2. HPE MSA 2060 Storage Array - Overview

Please click on "Thumbs Up/Kudo" icon to give a "Kudo".

Thank you for being a HPE valuable community member.

[Cali]

Hello,

yes, the MSA can do that, but only if it's configured and wired correctly.

Then it also needs to be configured on the host side, keyword: MPIO.

Since you haven't mentioned your operating system, I can't give you any specific advice.

But this guide is a good starting point: HPE MSA 2070 and HPE MSA 2072 Storage best practices

Cali

[idrissi]

thanks for your reply.

hosts  directly attached to msa (with cat6a cables). OS is vmware 8.0

[srujan_kumar]

HI @idrissi 

In addition to Cali Reply, path failover works via the built-in ESXi multipathing mechanism (VMware NMP), which switches to another available path using MPIO software and HPE MSA firmware, but this requires multiple independent paths to be available.

For that to work reliably with VMware ESXi + MSA 2060 iSCSI, the supported and tested architecture is the following:

• Two dedicated iSCSI switches (one per network/subnet)
• Each ESXi host has one vmkernel NIC connected to each switch
• Each switch connects to one MSA controller port (Controller A to Switch 1, Controller B to Switch 2)
• This gives each host two fully independent paths, one per controller

HPE only provides support for configurations listed within SPOCK, which is a database of tested interoperability between all active components, including the OS, HBAs, switching infrastructure, and arrays.

Recommended Path Selection Policy on VMware

If you do get into a supported topology, HPE recommends using the Round Robin path selection policy (PSP) for best performance. By default, VMware ESXi uses MRU (Most Recently Used), meaning it only uses one path until it fails, then switches. Round Robin actively load-balances across all available paths.

My Recommendation

1. Short term: If you must stay direct-attach, connect each host with two cables one to Controller A port and one to Controller B port. This at least gives you controller-level redundancy, even if HPE doesn't officially support it.
2. Long-term/proper fix: Introduce even one managed switch per iSCSI network. This is the only way to get a fully supported, reliable multipath configuration with VMware ESXi on the MSA 2060.
   
   Regards
   Srujan

[JonPaul]

Just to clarify,
HPE and VMware support direct connected iSCSI as of VMware 8.0
It is noted in the HPE SPOCK configuration for MSA 2060 iSCSI and VMware 8.0
"iSCSI Initiator Notes
1) All standard ProLiant NICs are supported in Switch Connected in conjunction with the iSCSI OS initiator (SW initiator). Direct Connect is supported with 1/10/25Gb Ethernet adapters."