Came across this Microsoft site with MS Triva type of questions for Windows 7 in enterprise environment. It is typical MS question you would see on their textbooks. Personally, I think it is a very good (and quick) way to learn more about Windows 7 if you haven't had much time to read the books yet.

http://www.microsoft.com/click/areyouce ... fault.aspx

PS: I only managed to score 55122 at the end of the test. Guess I should resign and move on to bake muffins for living.

I am oncall this weekend. Normally the afterhours work will come to oncall guy and this week, that's me, again. So nothing fancy, just got to keep the phone full of charge and get prepared to wake up by calls every now and then (which I am quite glad it didn't happen for the past 18 months).

So I was implementing a change request this morning. The change was about attaching 2 new luns to out existing file cluster servers. The shared disks were already presented and visible to the cluster nodes. I thought I just need to partition and format the disks according to our workplace standards and good practices, create 2 new physical disk cluster resources and all should be sweet.

The tasks went smoothly until I tried to do a failover tests. We always did the failover test on our cluster nodes whenever we had a software/hardware configuration change. A couple of the existing disk resources was "hung" in pending online mode for more than 20 minutes, I was panic.

My first step was to look at hints from the event log. System log showed only one error. The too general event 1066. Not exactly helpful for troubleshooting in a cold Saturday morning.



I spent a few minutes searching for the event related information. Within minutes I concluded that I need to run the chkdsk /F (as suggested by the event log, but how? The shared drives were not accessible from the Computer Management Console. I can only tried to set the dirty bit and hopefully the next restart it will run the chkdsk utility. ERmm.. and how exactly can I do that? I never encountered this before. Stuff it, I thought I might just try a few simple things and see how I went.

I then remove the newly create cluster disk resource from the cluster resource manager. Didn't help, but better. Now only 2 out of the existing 14 shared disks attached to the cluster node appear to be not responding. I then shutdown the second node of the cluster. Thinking that should force the first node seizing the ownership of the cluster resources. No help still. Time was running out.

F???, I thought, what can I lost. I restarted the last node in the file cluster. Something must had locked up the disks. Applying the Windows universal troubleshooting tricks (give the box a kick) should normally do the trick. 10 minutes later, hhhell no. Server came back up, 2 of the cluster disk resources still not coming online. Computer Management utility showed the disks correctly but not able to access it. The only clue I had from the system log 1066 still. #$%#^%$^

Time was running out. Change windows was almost reaching the 2 hours period. I started to call for help, at the same time also updated the status to a few stockholders of the workplace. After a few phone calls, I gathered a few things.

From what I had been told, it happen in the past to a few of the engineers too. Good, I was not alone. The bad news was, it may take as long as 2 hours for the disks to come online. No explanation was given. No one know why it was happen. It just did it. Well, at least I got some hope now. So what I needed to do, was waittttttttttt. And I did. I waited for nearly 45 minutes, staring at the monitor, and wwala, all resources finally came back up.

So I then booted up the second cluster node, thinking that since the resources were all functioning now, I should do a few failover tests again just to make sure they were really fine (and I was told that the long wait should only happen once). This time, I waited nearly 2 hours. Coincidentally, 2 of the disk resources holding up this time, one of them was not the one happen in the last failover.

Luckily Internet was working so I started to Google a bit further, while I was waiting for it to come online (at the same time I also picked up a Coke and a snack from the common fridge).

I see see look look, tried a few cluster CLI parameters to retrieve more information according to the links below:

http://msgoodies.blogspot.com/2005/04/h ... ng-on.html

http://support.microsoft.com/kb/223023

http://support.microsoft.com/Default.aspx?id=223023

http://msdn.microsoft.com/en-us/library ... S.85).aspx .

And I finally checked the one of the log files under c:\winnt\cluster consist of the output of the chkdsk log. I realized the cluster resource manager must be running the chkdsk to make sure the disk integrity. Wait, that explained the long wait in online pending. That also explained why it was doing to different disk resources. And I believe when I spent my first 5 minutes to research for the event 1066 I DID read about the cluster services log files, and I choose to ignore it initially. There was still mysteries however, why it was doing it to multiple disks this time, something must have triggered it.

It was all good after the second long wait. All of the resources were finally managed to failover between the 2 cluster nodes within reasonable time. I then recreated the new disk resources and did a few more tests. No more problem was found. So here I am. The chkdsk just made my day. This is not the end though. Tomorrow, I got an Oracle database cluster to deal with. Wish me luck.

So the MAC OSX look dock huh? I can still recall I was trying to get a similar feel in my Windows XP back to last year and it was awful. To set it up in Ubuntu I followed this instructions. All easy easy. After customized it a bit my desktop now looks like this. While, not bad for a few hours work huh?