Storeonce 5650 strange behaviour of remote support connection in case of links down

[sergio_rolandi]

Hello colleagues,

We are noticing, for the second time, on a 5650, a strange behaviour of the remote support connection.

During some availability tests we were shutting down a couple of 10GbE links of the 4 we have.

We have our 4 links configured in bond 4, LACP for both management and data, we don't have any 1GbE connection.

When our 2 links go down, also the connection to the remote support goes away, in 5650 dashboard telling us that we can't connect to the remote support.

When links are reestablished, the connection to remote support goes green again.

We thought that being in bond 4, the connection to the external site could and should go on every link.

It seems that it goes on a link and if this is down isn't able to switch to other links.

Can this be an explanation or may be a 5650 error or something else related to our environment that we could check?

Do someone else experienced the same behaviour?

As usual thanks and regards

Sergio

[Josh94K]

Hello @sergio_rolandi ! Good day!. 

Quick question !, Did you Try Bond Mode6 (Active Load Balance Bonding) instead of LACP to remove dependency on switch for NIC bonding?

@sergio_rolandi  I hope this helps!

If you feel this was helpful please click the KUDO! thumb below.

Note: "While I am an HPE Employee, all of my comments (whether noted or not), are my own and are not any official representation of the company."

[sergio_rolandi]

Hi Josh,

No, we didn't try it.

This is a big customer with 10 big Storeonce, 5650, 6500 and some old 6600.

Thay have always used bond 4 (don't know the reason why).

Before asking to switch to bond 6 I would like to know if our behaviuor is to be expected or not, then may be, I could ask to change it on the entire installed.

Logically speacking, using eth bonding, it should be a fault but.... who knows

thanks, best regards

Sergio

[Josh94K]

Hello @sergio_rolandi !
Thank you for your response,
Taking into consideration the scale of your infrastructure, it would be better to log a case with HPE so that they would be able to take a detailed look at your network/infrastructure topology and then just suggest whether to switch to bond 6 or not.

Regards
Josh. 

Note: "While I am an HPE Employee, all of my comments (whether noted or not), are my own and are not any official representation of the company."