I can’t believe it. Just when you think it’s safe to go back to reading the web… Part II of the worst clustering article of all time is out. Back in July (http://msmvps.com/blogs/clustering/archive/2005/07/21/58509.aspx) I wrote how bad Part I was. Several months later Part II came out and I missed it. A former ClusterHelp.com student asked what Russ and I thought of Part II.
Russ, summed it up nicely (http://msmvps.com/blogs/clusterhelp/archive/2006/06/23/102671.aspx) and I won’t restate what I wrote about Part I – even though it still fits. Let’s just say I hope there is not going to be a Part III.
Susan Bradley recently wrote about clustering SBS – http://msmvps.com/blogs/bradley/archive/2006/06/26/102877.aspx
Let me answer the question from a clustering point of view – you can’t do it. Plain and simple. If you truly need that level of High Availability – add a second Domain Controller.
I had an excellent time in Boston and truly enjoyed speaking at Tech Ed. What a great city to visit [:)] I brought my entire family along to enjoy it, my wife and 4 kids – even my Mother to watch the kids while the wife and I painted the town Red.
Looking back over my sessions I am overwhelmed by the turn out. If you combine the two clustering sessions approximately 15% of the Tech Ed attendees came to the sessions. That is simply amazing. It was a great feeling to stand in Grand Ballroom B and talk about a subject I am truly passionate about.
I also want to thank everyone for the kind words on the evals – I wonder if all the free t-shirts helped [:D]
I will hopefully see you next year in New Orleans!
I have lots planned for next week in Boston.
I am delevering 3 sessions, one is a repeat because of demand (http://msmvps.com/blogs/clustering/archive/2006/03/14/86357.aspx). Make sure to arrive early for the clustering session to ensure you get a seat.
I have been working with the Windows Server product team and I managed to get them to sponsor some give aways (two boxes full). I will not see what they sent until I arrive on the 10th to my hotel. It’s suppose to be logo stuff.
I can’t wait to attend the other clustering and High Availabilty sessions. Talk to the experts on the floor. Eat lots of great seafood. Post on comnet. See some old friends and make news ones.
I hope to meet lots of you too, so come on out and say hello. I will be the loud mouth guy with a NASA baseball hat on.
See you soon 🙂
Lots of advancements around our New York, New York training classes:
Every student now gets to use brand new Dell machines:
Pentium D Processor 820 with Dual Core Technology (2.8GHz, 800FSB)
2 GB DDR2 Non-ECC 667mhz SDRAM80 GB 7200RPM fast SATA serial drive
These machine are very fast 🙂 Yes, 64 bit rocks!
More dates have been added:
September and December have been posted.
Classes are starting to sell out:
The September class only has a few seats left, so book today to ensure you get a seat. If the demand warrants it Russ will fly out and we will run two rooms full 🙂
The December date (though just added) already has sold a few seats. Manhattan is wonderful just before Christmas. The air is simply electric.
LooksAlive: Will perform a cursory status check every five seconds (by default). This routine checks that the resource is not marked as “Failed,” which indicates a loss of the resource.
IsAlive: This is a much more complete check that occurs every 60 seconds (by default). This routine checks that the resource status is not marked as “Failed,” which indicates the resource failure. If the status is not marked as “Failed,” in the case of a Physical Disk resource a FindFirstFile runs on the root of the disk to make sure the file system is still mounted and that the disk is accessible.
Some resources are programmed to use both, some just LooksAlive. Ask your application vendor when troubleshooting Alive resources.
Verify that there are no dependencies or extra resources other than those created by the SQL Server setup on any of the SQL Server cluster resources. Resources such as file
shares, print servers, etc.must always have a unique network name and an IP address resource. You must remove any such non-default dependencies before you install SQL Server 2000 SP4. No dependencies other than those supplied by the SQL Server setup should be used.
For example, if you have MSDTC resource in the SQL group and if MSDTC is
dependent on SQL Server Network Name and SQL IP Address then you may have
this issue. If this is the case then you will need to remove the dependencies and then try running the setup again.
If you still have the same problem, then the next step will be to review
the logs. I would start with the sqlsp.log (created on the node where you
ran the SP4 setup).
SQL Server 2000 Service Pack 3, Service Pack 3a, or Service Pack 4
pre-installation checklist for SQL Server 2000 virtual server for failover
Please, please, please – only do the steps below IF you don’t have a cluster on said machine. Only if your machine thinks it’s a clustered node.
Try this first:
From the command prompt (where nodename is the machine’s name):
cluster node nodename /forcecleanup
Chances are that the following key and “ClusterInstallationState” DWORD
value under it are missing on your server:
If that does not work…
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\Current Version\Cluster Server
Create the key and the DWORD value by following the steps:
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\Current Version\
Right click –> New –> Key “Cluster Server“
Right click on “Cluster Server” key –> New –> DWORD Value
Name it as “ClusterInstallationState” and set it to 1
0 = ClusterInstallStateUnknown – Cluster Service configuration unknown
1 = ClusterInstallStateFilesCopied – Files Copied but Cluster Service not configured
2 = ClusterInstallStateConfigured – Cluster Service configured, fresh install
3 = ClusterInstallStateUpgraded – Cluster Service configured, upgrade
If you use the Print Migrator you may run into an issue, say if the location moved from F:\PrinterDrivers to P:\PrinterDrivers.
F:\PrinterDrivers” is current location
T:\PrinterDrivers” is desired location on shared disk
1. Check the following registry entry to confirm the path:
The above entry should be pointed to “T:\PrinterDrivers” ( shared disk) Note GUID maybe different, look for the ClusterDriverDirectory parameter.
2. If this is not pointing to “F:\PrinterDrivers”, stop the Cluster Service on BOTH nodes, and run this command from the command prompt :
cluster . res spooler /priv ClusterDriverDirectory=”T:\PrinterDrivers” <ENTER>
3. Restart the cluster service on both nodes.
4. Confirm that the path has remained fixed at T:\PrinterDrivers via the spooler resource and the registry entry.