Reset health issues

It seems to happens a lot that issues that are resolved still show up as critical in the Health Explorer. Using then Reset Health option doesn’t fix the problem though, but you can go to the Monitoring Pane / Operations Manager / Agent / Agent Health State /Agent State and select the agent of the system that is having the issue and use the ‘Flush Health Service State and Cache’ task. This will fix the problem most of the time. Annoying though.

SQL cluster gotcha’s – part two

After the installation of the OperationsManager database on the cluster, I removed my current Opsmgr setup (which was the everything-on-one-box one) and reinstalled it pointing the SDK to the instance on our SQL-cluster. I ended with upgrading to the SP1 RC.

Everything seemed fine, but nothing seemed to be monitored, and my opsmgr event log was getting weird messages like

A Bind Data Source in Management Group GW1 has posted items to the 
workflow, but has not received a response in 12722 seconds.  This 
indicates a performance or functional problem with the workflow.

This took us a while to find out, but the problem was caused by SQL. More precisely because the Service Principal Names for the SQL service weren’t set in the AD database, which caused the Windows Authentication in the SQL server connection to change to NTLM. Which in turn caused the Opsmgr SDK service to stall and puke out above mentioned errors.

We fixed it by using setspn

I didn’t find any notice of setspn and the above issues in the Microsoft documentation though (maybe I didn’t look hard enough 🙂

SQL cluster gotcha’s – part one

My first installation was an ‘everything-on-one-box’ one. Everything just worked fine, but not really scalable of course. So the powers that be decided to put every component on another server (still using only one management server though).

The Opsmgr OperationsManager database would be installed in a SQL cluster, so we used the Microsoft recommendation and copied dbcreatewizard.exe and a bunch of dll’s (from the SCOM directory!) over to the SQL server (very userfriendly there Microsoft).

Now the SQL server was running 64-bit windows .. and my Opsmgr server and dll’s were 32-bit. This didn’t seem to work very well on the SQL-server 😉 So we had to install Opsmgr on the SQL-server to copy the dll’s to run the dbcreatewizard to create the OperationsManager database. (and people complain about unix ..)

We now had a working database (or so we thought) ….

More to come in part two

Powershell automation – take one

I’m from a unix/linux background, so that obviously means I’m GUI-challenged. That’s why I’ve took it upon myself to try to manage everything in Opsmgr with Powershell.

So, what is one of the first things you do when you’ve installed Opsmgr ? That’s right, installing agents.

First a bit of background: I’ve chosen ‘local system’ for the server action account, and I’m using a domain account with local admin rights for installing the agents. This works fine with the GUI, but a bit too much clicking/typing for me.

So off to powershell I went, with the ‘install-agent’ cmdlet and using an example from the scomshell blog

$creds = Get-Credential
# Define a WindowsDiscoveryConfiguration
$dconf = New-WindowsDiscoveryConfiguration –ComputerName: "dc1.tad.loc" –PerformVerification: $true –ActionAccount:$creds -ComputerType: "Server"
# Start the discovery process.
$result = Start-Discovery –ManagementServer (get-managementserver) –WindowsDiscoveryConfiguration $dconf
# Check that the discovery process discovered the Windows computers you specified.
$result.CustomMonitoringObjects
# Last but not least install agents on the discovered computers.
Install-Agent –ManagementServer (get-managementserver) –AgentManagedComputer $result.CustomMonitoringObjects

Everything works fine until I get at the install-agent cmdlet, there’s no way of specifying any credentials, which means the cmdlet is trying to install the agents as ‘Local System’ which of course doesn’t work.

Anyone out there who has a solution for my problem ? or maybe I’m just doing something wrong, I’m from the unix side afterall.

Hello world!

The purpose of this blog is sharing my experiences with the System Center Operations Manager 2007 product.