3

We are trying to have an HA multipath where from 1 Server (Linux Fedora OS), we connect to 2 different storage devices (HPE 3Par Storage) that are synchronously replicated. Our idea is that in case one of them goes down, the Server can still access the information on the other storage.

It "kind of" works. But the filesystem mounts in read-only mode if write operations are being carried during the failure of the active storage. So we need to restart the Server and then it works fine.

Any idea if this is even possible to do? Any hint, link or document we could have a look at that point us to the right direction?

Thank you!

Improve this question

2 Answers 2

Reset to default
3

This requires a bit of technology outside the world of standard FC. Different vendors have different solutions. I know Hitachi, EMC, and Netapp all offer this, but it’s elaborate and involves, at least for EMC, client side software. I did a quick Google search and didn’t find anything for 3PAR.

That said, it’s not rocket science. If your server has the ability to validate the health of both storage devices, as well as make the replica read write, you could write a monitor that reacts to the loss of the primary device by making the secondary device writable and failing over to it.

Improve this answer
0

I am not sure if the connectivity from your server to the 3PAR storage is via ISCSI or FC.

Firstly, you may like to check if the Server OS and its version is supported with the 3PAR Hardware model and 3PAR OS version in your setup. If the connectivity is FC you may like to check that your FC HBA model and FC HBA driver/firmware version on the server is supported for connectivity to the 3PAR array.

AFAIK, RedHat Enterprise Linux 7.x, 8.x and 9.x are indeed supported with the HPE 3PAR. The earlier storage hardware models were branded as "StorServ", now it's branded as "Alletra" i think. The brand name for the storage replication across 2 or more 3PAR systems is now called the "Peer Persistence" feature.

There are some other things you may like to check from the 3PAR Linux Implementation Guide:

  1. That the Volume from the 3PAR storage mapped to the Linux Host has the right Host Persona configured for it. In particular, the Host Persona as recommended by HPE for 3PAR replicated configurations with a Linux server.
  2. That the multipath settings on the host are as per the recommended settings from HPE for 3PAR replicated configurations. The DM-multipath settings are required to be configured for ALUA persona. Also, the dev_loss_tmo value in multipath.conf must be set as infinity. The value of infinity is a required parameter for replicated configurations.

Refer:

  1. HPE SPOCK website (Single Point of Connectivity Knowledge for HPE Storage Products)
  2. HPE Support Center website
Improve this answer

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .