HPE MSA 2050 iSCSI problems

[IT-Dreamer]

I hope someone can help!

We have a HPE MSA 2050 configured for iSCSI with 4 ports enabled, 2 from each controller.

These connect into a bridge aggreation group on a HPE FF 5700.

We also have 5 HPE GEN10 servers, without enought ports (that's another story).

We have 1 iSCSI fibre port connected from each ESXi 6.7 server, through the FF 5700 and into the SAN.

It connects via it's own VLAN, the ESXi server has it's own Vkernal, Vswitch with an IP assgined and the VLAN and ports of the SAN are in the same subnet.  The SAN is in a seperate subnet for management.

We add the dynamic targets in ESXi and the static targets are populated.

The SAN has the host, initiators and mappings all setup.

Here's the problem:

The SAN, in the Discovered colummn, will change from Yes to No and the ESXi servers lose their datastores, that sit on the SAN.  This seems to be random, if we restart an ESX host another server can drop it's connection to the SAN.

The IP traffic on the Vlan has Jumbo frames enabled at 8900, I have done vmkpings and confirmed that I get a response.

Please help.

Thanks in advance.

[HPSDMike · Accepted solution]

Well, off the top, I would say that the MSA host ports don’t support the use of link aggregation. Instead, each MSA port gets an IP address and load balancing occurs via the host OS multipathing configuration.

Take a look at the best practice guide which discusses how host ports are typically configured and multipathing is set up.


https://h20195.www2.hpe.com/v2/getpdf.aspx/a00015961enw.pdf

[IT-Dreamer]

Great, thanks, I've been going round and round in circles trying to solve this.

[EHANN]

Hi,

Im having the exact same issue with same hardware and model.

In the end, how did you manage to fix it ?

Thanks for your help

[IT-Dreamer]

Hi

I followed the advice of the solution and didn't use BAGs in my switch config