HPE Storage Users Group

So we never have any luck with Inform OS upgrades. Seems with every version they make changes to the NIC drivers. We have had to flip flop between our management ports being set to AUTO and then locked to 100 FULL with every release. Our RC ports had similar issues in 3.1.3 such that we were locked out of the entire array with no RC until a tech could come on site and serial into the array. Also in 3.2.1 we are finding issues with AO tasks failing. And since going to 3.2.1 one of our arrays shows unknown status int he new SSMC console, but is fine in IMC.

Don't get me wrong, I understand the complexity of code, but 3par QA is suspect here as well as their inability to react when we encounter these issues. I get the "deer in the headlights" reaction from support and have to leverage my account team to escalate cases to make even the smallest progress. If it were not for flash cache (which we are not eve sure will benefit us) and express writes I would stay on 3.1.3 as long as possible as I am so tired of these painful upgrades.

I can imagine you are sick and tired of it!

It seems like every update since 3.1.2 has given you major fits!

It just goes to show that scale means a lot. I have a small-ish environment (1 vmware cluster per site, 2 sites, 2x 7200's @ 60TB each), and the upgrades from 3.1.2 MU2 to 3.1.3 MU1 and from there to 3.2.1 MU1 have been flawless.

I'll admit that I haven't upgraded to 3.2.1 MU2 yet because rule A) I enjoy spending my evenings at home with my family and rule B) unless there is a good reason, I will not violate rule A.

and as best I can see, I gain nothing other than some overtime pay from doing 3.2.1 MU2

If it were not for Flash cache and Express writes I would not have. I have no plans to go further, unfortunately the new SR will require at least a patch to 3.2.1. What amazes me is we use Cisco switches it is not like we are using obscure switch hardware.

We are pretty new with 3PAR's. We've had some for over a year now but we are starting to get more. Right now we have eight arrays. All the upgrades we have done went rather well. However, the upgrade to 3.2.1 mu2 caused all our arrays to pause during the upgrade. Granted no connectivity was lost but at a certain point in the upgrade, the array mgmt port went offline. The reason seemed to be that during this particular upgrade the cisco switches marked the ports as offline due to all the flapping. A simple reset of the mgmt ports on the switch side (all of them) got the arrays back on the network and it finished without any other issues.

As you can tell I found this quite odd. I asked the networking guys if this was a new feature they implemented but they said it is standard on all their switches. We are plugging into nexus 2000s into 5000's into 7000's. Anyway, my network guy for a while has been telling me that the active port for the 3par looks good but all the other ports not being used are constantly flapping and filling up the logs. I will be opening a case with HP for this when I get around to it. It isn't a high priority right now.

So yes...we saw some networking weirdness with mu2 as well.

the unused network ports flapping like that has been around for quite awhile.

If you search the forums you'll find a few threads describing it. Unfortunately the (admittedly pat) answer from HP is that it's expected behavior, and you should configure your switches accordingly.

The internal health check brings the secondary mgmt ports online every once in a while, checks if everything is ok and then brings it offline again. The port should not be flapping in a rate that would cause any issues. The flappig during the upgrade to 3.2.1MU2 sounds like a seperate issue, no idea what would cause the rapid port flaps during the upgrade though.

I know of a few customers where support somehow disabled the healthcheck on the secondary ports. it stops the errors in the switchlogs, but the ports then go untested, which might lock you out of the array during a network failover.

_________________
The goal is to achieve the best results by following the clients wishes. If they want to have a house build upside down standing on its chimney, it's up to you to figure out how do it, while still making it usable.

Evidently HP recognizes this is a bug and is issuing a patch 3.2.1 MU2 P11 will address the network issues.

we experienced loss of connection to the node management ports whilst updating to 3.2.1 MU2. when updating the second site our network guys changed something on the switches so it didnt happen the second time.

did they provide any ETA on that?

_________________
The goal is to achieve the best results by following the clients wishes. If they want to have a house build upside down standing on its chimney, it's up to you to figure out how do it, while still making it usable.

They did not, but my guess would be within the next month or so.