Manager error message

[Peter J West]

Hi,

I'm getting an error in the CMC when logging in.  Anyone got any clues on how to fix this one:

The following systems report that they are running managers in a management group, but the management group thinks they are not running managers:

It then gives me a web address which to be honest is no help at all as it's just the address for the support portal.

Is there a procedure to fix this?

Thanks

Pete

[Bart_Heungens]

Which version of software are you running?

How many nodes? Is there a FOM?

What happens when you manually start or stop managers on some nodes?

[Peter J West]

Hi Bart - again ;)

The problem seems to have developed somewhat - now the name of the node in the cluster has changed to just be the IP address and clicking to expand it says 'Could not find storage system with serial number RIP0-[MAC ADDRESS]'.

The node is appearing further up the list inside the management group but it doesn't look like it's part of the cluster any more.

The status says:

Storage System not ready: No Devices, Storage System No Device Groups

The feature registration tab says it's not licensed even though the licence key is correct.

I'm running Build 0163 v9.0.0.1 of the software and this problem has really confused me - but how do I recover from it?

Thanks

Pete

[Peter J West]

Update:

I did a 'Repair Storage System' and chose 'Storage System'.  Now the node is not in the management group but the status of it looks ok - does this mean I should be ok to try adding it back into the MG again?

Or will it cause problems because the old 'broken' entry is still showing as part of the cluster?

Thanks

Pete

[Bart_Heungens]

 I never saw this problem before so I only cqn say what I should do in this case...

What worries me is that, if you add that node back in the MG, it will loose all data on it and data from the existing MG and CL will be sync'ed on it...

Are your volumes still online on the other nodes?

[Peter J West]

Yes all the other volumes and nodes are fine - and the SAN volumes themselves are working because the affected node is part of the second site.

I tried adding to the MG again and it eventually came up saying it needed to be registered.  :(

[Bart_Heungens]

Registration is based on the MAC address of the first NIC... Does this mean the MAC address changed from the node? Seems not realistic...

[Peter J West]

Hi Bart,

Sorry - I just checked and the MAC looks the same?

Strange.

Pete

[Bart_Heungens]

When you create a bond normally it takes the MAC address from the first NIC, but off course for Linux (and also Windows and ESX) what is the first NIC ? :smileyhappy:

[Peter J West]

I'll remove the bond and see if it will add to the MG then.

Edit: I tried and still the problem is the same.

[Peter J West]

Ok - i'm not sure if this is related but i'm going to do a test to find out.

The node with the issue only has a single NIC enabled at the moment and it has a static IP - but the MAC of the NIC is different to the feature ID shown in the registration page.

I'm wondering if it will work if I switch the IP to the other NIC (the MAC of which matches the Feature ID).

I'll report back shortly.

[Peter J West]

Ok, I think i'm finally making some progress but it's been a bit of a trial.

Here's what i've done.

1. I ended up using an ILO connection and forcefully removing the node from the Management Group. 

2. I then used the same connection to reset the node back to it's default.

3. Next I reconfigured the interfaces on the node so that interface #1 had the correct IP address.

4. Then I entered the license details again.

5. I tried joining the node to the management group but it complained about not being licensed.

6. I ignored the error and edited the cluster - it then let me exchange out the unit which wasn't responding in the UI.

The cluster is now showing no errors but of course I have to wait for it to restripe and resync across the cluster.

Only one problem now remains - the old node which gives the error 'Could not find storage system with serial number RIP0-MACADDRESS' is still in the management group.  Worse of all the node appears in the UI with the IP of the node which is actually the CORRECT IP for the node i've just added in, so i'm a bit worried that trying to remove it could result in the working node also being removed.

For now i'm going to leave it - i'll wait until the restripe has completed before proceeding.

[Bart_Heungens]

Correct first wait the re-syncing...

Normally the CMC detects the nodes with the MAC addresses so if the MAC address changed, you can remove the other node inside the CMC...

[Peter J West]

The resync finished and the node which was invalid automatically dropped out of the management group and vanished.

After being patient things have finally settled down.

I still need to enable flow control and jumbo frames but i'll leave that task for another day I think. ;)

Again, thanks for your help

.

Pete

[zaragozam]

I had exactly the same problem. I resolved it by disabling jumbo frames. (Thanks to another HP post in the web)

In my case I was setting up a new P4300G2 , 2-node cluster.

So, now the CMC allowed me to create the Management Group.

I wonder if it would be possible to enable Jumbo Frames now the Management group has been built.

I will let you know.

[jobecker]

I had to shut down a VSA on a VMware-Server for maintanance-mode. After rebooting the VSA the centralized management console could not reconnect the storage. 

When I tried to reboot the VSA from the cmc it said "Could not find storage system with serial number...".

I checked the mac-address and it was okay.

At least I started a search for VSAs from the cmc -> find -> find systems.

After this the cmc could reconnect and the error was gone.