Multiple "link up" messages, SP 8.25A, Windows 2008

[suzan maxey]

I recently built the first Windows 2008 (standard) server in our domain and installed SP 8.25A. During reboots of the server, I receive a normal series of emails from my Insigh Manager server except for a deluge of about a dozen "link up" messages at the end of the boot process. Any insight as to what is going on (and hints on how to fix it!) would be appreciated. Thanks.

[John Benak]

Got the same problem. Rolling out 2K8 in our enterprise and the same thing happens. Only thing I've been able to ID so far is that looking at the server in HP Sim 'data collection' page shows about 12 NICs, each with the same MAC addy. We're thinking it might be something to do with the redesigned 2008 network stack, and HP Sim is seeing them as separate NICs as opposed to the same one. BTW - this is an ESX Guest VM, and the problem is replicated on all. Please let me know if you come up with anything.

[suzan maxey]

I also noted the multiple NICs in the data report, went to the server, and disabled everything except IPv4. Still the same result on reboot: a torrent of LinkUp messages. I rebuilt the server with a different name, installed an earlier SP, hoping that something had gotten screwed up the first time. Same result, though.

I finally ended up going to Insight Manager and disabling the LinkUp trap. (Options -> Events->SNMP trap settings.) the trap is in mib rfc1215. I still get enough traps on reboots to tell what is going on (coldstart, etc) but it keeps my mailbox from being flooded with LinkUp traps. Not the optimum solution but that's the best I could come up with.

[John Benak]

Thanks for the info. Disabling the 'link up' component isn't really a solution for us. I'll look around a little more and then call HP support if I don't find anything. If I get a solution from them I'll report it here.

[CNUXX]

I also have the same problem. Were you able to get a resolution/explanation from HP?

[John Benak]

I worked with HPSim support yesterday, and we were able to determine that this isn't a problem with HPSim per se, so much as something with SNMP in Win2008 (or SNMP in 2008 on VMware guest). We confirmed that by runningg SMNPUtil on our HPSim box, rebooting the 2008 server, and watching the raw traps coming in from the test 2008 box. HPSim was only reporting what it was getting.

What I'm not sure yet is if this is specific to ESX VM Guests, or whether this will be the case with physical servers as well. All of our 2K8's are ESX VM's right now, but I'm in the process of building out a physical server to see if there's any difference in the trap reports. The VM guests don't have the agent installed, so that may be a factor as well.

Has anyone seen these errors on physical servers? Please let me know what your experiences have been with this.

Google hasn't been too helpful. If I can get some info on your experiences/scenarios, I'll go to either MS PSS or VMWare support next and post results.
Thanks,
John

[suzan maxey]

The server I'm having issues with is a physical server -- Proliant DL360 G6 -- so this isn't limited to VMs.

[CNUXX]

I am having the problem on a DL380 G6 running Windows Server 2008 SP2 with the Hyper-V role enabled. Perhaps each HP agent and/or service is sending a message that it has started. I noticed that agent name is missing. Additionally are all of you using SNMP instead of WBEM?

A linkUp trap signifies that the sending protocol entity recognizes that one of the communication links represented in the agent``s configuration has come up.

[John Benak]

Not sure about WBEM, as I didn't build the HPSim infrastructure. Can you enlighten me about it? As for SNMP, we're certainly using it.

To illustrate what I'm talking about with my issue, take a look at a 2K3 server and a 2K8 server. Go to the 'Tools and Links" tab for that server and run the 'Data Collection Report'. In my case, comparing 2 VMs in this report I get significantly different results, notably in the network/NIC section. Where there is only 1 entry in the 2K3 report, there are about 15 in the 2K8 one - most reporting the same MAC addy.

I think this has something to do with the new 2008 network stack/config. SNMP is reporting 'linkup' for each protocol, instead of just for the NIC/IP. What I'd like to be able to do is turn off the firehose of info.

[John Benak]

I opened a MS PSS ticket this AM, during which I clearly and undisputedly demonstrated & illustrated the nature of this problem via the HPSim interface and the shower of notifications during a 2K8 reboot. Got a "That's interesting... let me put you on hold...".

After having checked things out with a Escalation Tech, I was told the following by PSS:


"As we discussed, it appears that Windows Server 2008 is sending traps on the additional interfaces (ipv6, wan-mini, qos, etc) where 2003 did not. I will dig into the captures you provided. Unfortunately, if this is the case, there is currently no configuration granularity within the available SNMP interfaces that would allow you specify which interfaces to alert on. The solution may be to turn of Physical alerting, as being suggested in the HP threads, and using EVNTWin to alert on selected Event Log entries, such a start up events or service start event."

The latter suggestion is the use of a little-known utility to create custom alerts. Not sure how efficient this will be in a large enterprise.

I sent him network captures that the HPSim box received during a 2008 server reboot, so it's not TOTALLY written off, but it doesn't look like this is configurable from the 2008 OS perspective.

HP is coming in a week or so to upgrade our HPSim to the most recent version. Is anyone running the most recent, and is there any way to trim the alerts at the HPSim point in this version?

If PSS comes back with anything else, I will post results here.