A Fish

Travelling back from a customer I remembered an old saying.

Give a man a fish and you feed him for a day.
Teach a man to fish and you feed him for a lifetime.

I was thinking about that in relation to Ops Manager and I have come up with.

Installing OpsMgr keeps you busy for a day.
Tuning, tweaking and getting it to do exactly what you want keeps you busy for a lifetime.

This came about as a number of customers think that the installation is the main job of the project when in reality it is just the start. I can see why they might think that if they have done AD and Exchange projects. In those type of projects there is an awful lot of planning and preparation but in general once it is installed it can be handed over to operations to look after on a day to day basis as they will be doing business as usual – creating accounts and checking all is working. Ops Manager does not require anything like that much up front work but it does take a lot of work after it is installed to match it to the organisations requirements. Even after you have tuned it they then come up with all sorts of ideas for custom management packs and reports. Then new MPs are released which usually bring a whole new set of foibles with them.

(And more complex setups will take more than a day to install but that did not fit in with the fish saying)

R2 Info Starts to Flow

I have downloaded OpsMgr R2 but I have not had time to install it yet. But plenty of others have. So if you are looking for some quick info here are some links.

Robert Smit provides screen shots of the install (looks the same as standard 2007 except for the requirement of ASP.NET Ajax Extensions 1.0 for ASP.NET 2.0 plus you need to remember UAC if doing it on Windows 2008)
http://www.aca-computers.nl/?p=157

Robert follows up with screen shots from the Cross Platform Extensions (now part of the product)
http://www.aca-computers.nl/?p=158

Anders Bengtsson has picked on some of the new stuff with screen shots at
http://contoso.se/blog/?p=304

Walter Eikenboom has tried installing it with Windows and SQL 2008 to prove that it all works.
http://weblogwally.spaces.live.com/Blog/cns!A913F865098E0556!524.entry

UK MVP Gordon McKenna provides a very positive and incredibly enthusiastic appraisal of the product at 
http://wmug.co.uk/blogs/gordons_blog/archive/2008/11/27/operations-manager-r2-totally-unleashed.aspx

Well it would not be Gordon otherwise. Gordon also posts about the new forums for the Xplat bits f R2.

http://wmug.co.uk/blogs/gordons_blog/archive/2008/11/27/new-opsmgr-cross-plat-amp-interop-forums.aspx

If you missed the SCVUG meeting on Friday you missed a great session on R2 by Baelson Duque from the Product Group. The PDF of the session is at
http://www.systemcenterusergroup.com/files/folders/meetingnotes/entry227.aspx

You need to login (free) to SCVUG. You can get the Live Meeting (32 MB) at Skydrive
http://cid-1590b798c9cd6d68.skydrive.live.com/browse.aspx/Public/SCVUG/Nov2008?uc=1
but they do mention that “the quality of the Live Meeting video is not great (so demos are grainy), but the decks and audio are good”. They are not kidding. It is almost unwatchable. Shame as the demos were what made the presentation for me.

But is not all light and sweetness as Marius Sutara blogs on an issue with computer group and dependency health rollup in R2. Well it is a beta!

http://blogs.msdn.com/mariussutara/archive/2008/11/26/computer-group-and-dependency-health-rollup-in-opsmgr-2007-r2-beta-release.aspx

Single Server View

The other month I wrote about views and how to use them to help focus on particular areas. One of the views that I sometimes want to create is for a single server to get a feel for all the alerts that are being generated by that server. In MOM 2005 it was trivial to create a view like that but with 2007 it is much harder with the object orientated nature of the product. You can, however, use the server name as a filter for a State view but not an Alert view.

Here are some ways I look at getting around the problem.

1 – Create a view using “Raised by an instance with a specific name”

2 – Create a view using the server name as part of the search “With specific text in the description”

3 – Create a computer group with that server as the only member (in fact I searched for the computer name and put all objects that were returned in the computer group)  and create a view based on Entity but filtered with that computer group

 

The first one is closest to filtering on server name but it is not 100% as not all sources include the server name and in my test that server showed 12 alerts. The second one works better than I thought as even though the alert description does not appear to contain the server name it still picks up alerts from that server but still only came back with 9 alerts. The last one was the most comprehensive and picked up alerts from both other views (20 alerts) but it is a bit of a pain to create a computer group just for a single server. Perhaps that is something that a future version could do. Automatically create a computer group for each server and keep it up to date with all the relevant objects for that server.

In the end I suddenly realised that I could search on the list of alerts for that computer! And that came back with 20 alerts. So it looks like the easiest and simplest way to get a list of alerts from a single server is to use the server name in the search box for an alert view. However when I used the same server name in the search box in the main Active Alerts view rather than the one I had created in my workspace it only returned 18 alerts although in essence the views were the same. I did change it so it was exactly the same but my alert view still showed 20 alerts to the main Alert View showing 18. The main view did not pick up alerts where the source was sysvol or netlogon whereas the view I had created to mimic that view with the similar properties (just personalised differently) did pick those 2 alerts up. Very strange. I guess the moral is be careful when creating views as what you use to filter may filter more than you want.

OpsMgr SP2 Beta 1 on Connect

If you have been reading any other OpsMgr blogs you will know this by now but Beta 1 of OpsMgr SP2 is now available from Connect. No mention of a SCE version.

There are three downloads.  A highlights document (Word – 69 pages – if that is the highlights what is the full document going to be like! Most of it appears to be about Xplat), release notes (HTML) and the download exe which is 1,126.42 MB. If you are downloading this one make sure you have a fast connection!

Some bits that have intrigued me from the highlights:-

You can browse the management pack catalog directly from the wizard and import it directly or save it to a file.
A new template you can use to easily create process monitors.
In the Windows Service management pack template, you can use wildcards when you specify the service name that you want to monitor.
The Health Explorer has been added to the Web console. (Much need if you use the web console as that is the best way to find information on monitors.)
You can install the Operations Manager 2007 R2 databases on SQL Server 2008.
When you create a dashboard view a search tool helps find the views you want quickly. In addition, the search also includes views that you have created in your My Workspace. (That last one is great as that has frustrated me before.)
You can put a whole computer into Maintenance mode. This automatically puts the Health Service and the associated Health Service watcher into maintenance mode, which suppresses all alerts on that computer. (Hurrah!)
In Operations Manager 2007 R2, the Operations console performance has been greatly improved. These improvements are most evident in: 
• Opening new views in the monitoring space
• Pivoting between views
• Selecting multiple items in the results view and rendering the details pane more quickly
• Indicating that a view is in the progress of loading
(Well we will see about those! But any improvement in performance is greatly welcomed.)
The notification feature has been restructured to make it easier to configure.
You can create a new subscription directly from an alert and to add the parameters from the selected alert to an existing subscription. (This will save messing around with all those PowerShell solutions.)
Support has been added to natively monitor IIS 7 without having to enable the backward compatibility APIs or legacy management features
Operations Manager 2007 R2 supports monitoring as many as 2,000 URL monitors per management server.
The new version of the authoring console adds several user interface enhancements, and now includes the ability to edit any element in a management pack.
(And one that I have blogged (complained) about in April and Kevin Holman has blooged about. ) Reporting – it is easier to find the objects that you need to customize a report. The object picker has been enhanced to enable search and filtering by name or class of object, which makes it easier to find the objects.

So I now need to find the time to download the beta and get it installed and see if all the promises above are worth the upgrade. This is a beta so there will be bugs and performance improvements are generally done near the release but it certainly looks interesting.

And as I mentioned before it is not a free upgrade unless you have Software Assuarnce. And to confirm that they say on the site:-

“R2 is a new version of Operations Manager 2007 and it is not a free upgrade”

Server Management Magazine Article

I have had an article printed in Server Management Magazine. For those that don’t receive it you can read the article here:

http://www.server-management.co.uk/features/782

It is about getting started with custom reporting in OpsMgr. I go through a couple of examples of creating custom reports – percentage disk space and how to create a report listing all the HP ILO IP addresses in an enterprise which I have been asked to do a few times for customers.

Although I have written quite a few blog posts now it is still amazing to see what I have written printed in black and white. Especially in a magazine that I have been reading for years.

Disappearing DNS Community Fix

I mentioned the problem that a number of people are having with the latest DNS MP in  a previous post. Apparently Microsoft are aware of the bug but they can not repro it. Without being able to reproduce it they are having a problem fixing it. However Daniele Grandini contacted me with a fix that he has created. In his environment his DNS servers went from 38 to 53 after applying his fix and they have not disappeared since.

He says

The discovery script has a bug when the WMI query fails, in this case it will return an empty discovery document, this in turn will delete the previous discovery. This explains the flip flop behaviour with disappearing and reappearing DNS Servers. It must be added that the discovery runs every 15 mins making this more visible.

The repro is actually quite easy, if the following line fails the discovery returns an empty discovery document:
Set colServers = ExecuteWMIQuery(“Select * from MicrosoftDNS_Server”)

On one side is a WMI related issue (DNS provider?) on the other side the script doesn’t test for an error. To simulate such an error just enter an error in the line above, like:
Set colServers = ExecuteWMIQuery(“Select * from MicrosoftDNS_Server123″)

My mod is as least invasive as I can, at the end of the main routine I just checked for an error and in that case I won’t return any discovery data, it was:
                Call objAPI.Return(objDiscoveryData)
                Trace “Exiting normally”
Now it is:
If Err.Number <> 0 Then
    trace “Unable to CreateDiscoveryData.”
    wscript.Quit
else
                Call objAPI.Return(objDiscoveryData)
                Trace “Exiting normally”
end if

 

I have not had a chance to test this MP myself yet as I have not been back to the customer where I installed (and then uninstalled) the latest DNS MP. However it is worth a look in a test environment for all the people having this issue.

Here is the MP: – microsoftwindowsdnsserver2003fixxml - remove the .doc ending (needed to get it to upload) to get the XML file.

New WSS 3 MP v6.0.6447.0

Here are the changes for the MP from the guide. You only get this information after you have downloaded the MP and installed it to get the guide!

The following list includes all of the changes contained in version 6.0.6447.0 of the Windows SharePoint Services 3.0 Management Pack:
•    Updated the GetServerName.vbs discovery script that would previously generate alerts on any non-SharePoint server.
•    Changed classes to public, allowing monitoring to be added and the classes to be used in diagram views.
•    Updated the GetServerName.vbs discovery script so that it works on non-English operating system versions.
•    The SharePoint discovery scripts now use the Privileged Monitoring Account. This enables the management pack to be implemented in low-privilege environments. For more information, see the Low-Privilege Environments section.
•    Changed monitors to public, allowing diagnostics and recoveries to be added to the monitors.
•    Updated various display strings in the management pack.
•    Changed the health roll-up of the Microsoft.Windows.SharePoint.Services.3.0.Server.EntityState monitor to roll up to Availability state instead of EntityState.
•    Changed alert views in the management pack to look for non-closed alerts.
•    Changed the alert priority of all generated alerts to 1.
•    Changed the time-outs of the GetServerName and RecycleAppPool scripts to 300 seconds.
•    Updated knowledge for events 6398, 5586 and 3355.

Nice to know that some of these issues are fixed as the discovery one has been around for a while and the OpsManJam site had an “unofficial” fix out for that one for a while.

Looking at this list and the recent posts on problems with the DNS and ISA MPs it makes me wonder what sort of quality testing these MPs go through. Admittedly that some of these MPs can be quite complex but I would have thought that discovering the components that you are supposed to be monitoring would be key.

Ops Manager R2 not SP2

The announcements are coming out of the Teched at Barcelona and the word is that the next version of Ops Manager will be R2 and released about Q2 of 2009. Initially there was a lack of detail about whether this would be an SP2 release or R2 release. I had seen both mentioned. So what is the difference? An SP is supposed to be just that. A service pack that rolls up fixes to a known state and is free. And at one stage Microsoft were adamant that SPs would only be fixes but they have blew that several times since that announcement and many service packs contain updates and new functionality as well as fixes.

An R2 release is known as a minor release. The Windows server team would show slides of major releases of the OS every 4 years with minor releases in between at 2 years. The big difference is that SPs are free but R2 is for Software Assurance (SA) customers only. So if you already have Ops Manager but did not buy it with SA you can not upgrade to R2 for free. You have to pay for the product again. If you are buying new you would obviously buy it at R2 when it is released.

The beta should be available at the end of this month.
http://blogs.technet.com/systemcenter/archive/2008/11/03/operations-manager-2007-r2-beta-announced.aspx (includes link to video with Barry Shilmover talking about Xplat)

One of the big things that will come in this release is the Unix and Linux monitoring that is in beta at the moment. I have written about this before and I have been impressed by this but it looks like you will not be able to use it without having SA. They say there are improvements to the SLA Reporting and the Authoring Console. As those 2 are available already as free downloads I am not sure how they will fit into an R2 release as I would definitely expect the Author Console to be freely available to everyone. It certainly needs an upgrade as the current v1 product is very basic.

And I like the comment

enhanced system and web monitoring capabilities, and more

Well I would expect enhancements for a new release but “more” is wonderfully vaugue. I guess we will just have to wait and see.