MSCS Cluster monitoring with HPSIM v5.0 SP2

[Thomas Harder]

We just recently upgraded to HPSIM v5.0 SP2 and can't seem to get our MSCS clusters to show up within the MSCS cluster view.

I've verified the Clustering Information agent is set to being active (this shows up under Control Panel\HP Management Agents), the SNMP community strings are set correctly, the servers making up the cluster have been pulled from HPSIM and readded, the cluster alias name has been added manually but doesn't show up within the MSCS Clusters view......any ideas would be appreciated.

Can't really find any documentation on what needs to be done in order to make the clusters show up within HPSIM.

[Ferry Steenvoorde]

I've got the same "problem", but maybe I have some data for you that can help you further.

The cluster configurations (W2k and W2k3) I manage have at least two network interfaces: one for the Management Network and one for Production. Both networks are separated. The HPSIM-server is obviously on the Management Network and does its Automatic Discovery based on the IP-ranges on the Management Network.

At the moment I can identify one W2k-cluster. It has its cluster IP-address on the Management Network. All of the others (most W2k3) are not found, but they have their cluster IP-address on the Production Network that is not scanned. So I guess the cluster IP-address must be scanned in order to be recognised in HPSIM.
Added to that, IP-traffic from and to the HPSIM-server must be routed correctly (default gateways and subnet masks set correctly).

The HP Management Agents on the cluster servers are version 7.40; all of them got theirt Management Agents also monitor clusters.
The Systems Management Homepages on the servers that are part of the cluster show the cluster statusses and statistics correctly.

Hope this helps.

[Brian Rickett]

Tom, I had similar MSCS discovery issues with my updrade to SIM v5.0. I have several MSCS SQL Servers in our datacenter. Sounds like you've covered the basics. I will assume that you beleive your clusters' agents are working fine. Can you send a trap to verify? Did they work previuosly? Did you install the agents on each physical node after the cluster service installation? This may be the root cause. I have learned that the agents most times do not pick up the cluster services when installed prior to the cluster service install. I verify that by looking at the registry. You should see two additional keys for cluster extensions. Also resart the SNMP & Hp agents on each.

Ensure the following.

1. Verify the reverse lookup of all three addresses (Cluster virtual name & both physical nodes)work with nslookup from your SIM Console. Within our clusters' configuration I have enabled "Verify DNS registration succeeds" on the cluster IP and other IP address resources. By default it is not enabled when the resource is created. This will ensure that your Virtual IPs are in DNS.

2. Within Insight Console verify that Auto discovery is configured to recieve from that subnet. The upgrade may have affected this. I had to reconfigure ours. Complete ranges were wiped out. Our setup has several "Includes & Exclude" ranges. We never run "autodiscover jobs". We are configured to Discover upon reciept of a trap. The Nodes/Systems are auto discovered once the agents are comunicating correctly.

3. On the SIM console from the command line: Delete all three IP addresses (even if you do not see them!). Wait about 10 minutes. Believe it or not you may need to reboot the SIM console server. I have seen issues where deletes just did not happen and a reboot cleared it (SQL included! A stored proc might be locked).

4. If the physical nodes get Auto-Discovered when a trap is sent, manually add the Virtual name. I normally do not send a test trap at this phase of discovery. I use the command line tools provided with SIM to do IP address deletes and adds. The command line tools vs the GUI in our environment seems to work better.

We also had issues with the SIM certificates. So ensure that's not an issue as well. That's another discussion. So, those are the steps I always run through to get things working and I hope these suggestions make you successful.

Brian

[Etienne Beureux]

Same problem here, I found the JSP line that was causing the problem, but couldn't correct it (since it's a call to some class code).

I reported this on the forum at the end of October, but didn't hear anything from HP...

--------------------------------

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=961138

I did some "printf" debugging, and it seems the page "Systems Insight Manager\hpwebadmin\webapps\ROOT\mxportal\clustermonitor\CmMSCS.jsp" gets stuck when proceeding line 146 :
mscsTableModel.loadTable(loginBean,resObj,szSelected);

When commenting this line, the rest of the code is processed, and something loads (of course, commenting the line breaks the actual functionality of the page).

I don't know anything about JSP, though, so I can't fix the actual code.

-----------------------------------

[Phil Slator]

I had a similar issue with my fresh installation of SIM 5 SP2. It would recognise that two of the cluster servers were part of a cluster, but picked up the name of another node as the cluster name, and wouldn't allow me to add the other two nodes.
I deleted all of the cluster nodes from SIM, then loaded the PSP 7.3 onto the cluster servers, and manually added the physical server names to SIM. (I loaded 7.4 on one of the cluster nodes, but it started running very slow).

After leaving it to fester overnight, the SIM recognised all of the cluster resources properly.