LOUD FANS!!!

I have wrestled with the "LOUD FANS" problem in two different installations. One installation was a rackmount intel chassis/intel S1200 motherboard. The other was a pedestal intel chassis/intel S2600 motherboard. There have been multiple incidents between the two installations that I manage. Across these incidents, I have quieted the fans with 3 different actions:

1- total powerdown/go dark. Remove redundant power supply modules, count to 43(or maybe it was 44, you get the idea) replug power supply modules and go to town.

2- rewrite of BIOS/FRU/SDR/ME- this is an intel recommendation...upgrade of BIOS/FRU/SDR/ME or, if you're already on the latest, rewrite them.

3- OK, this is the one that got me tonight. I go to a job to upgrade RAM, I do my thing and the fans go nuts- LOUD like you're at the airport! Well, guess what? I hadn't replaced the stupid cover on the pedestal chassis! There are "contacts" there, and when I replaced the cover, everything settled down!

Just thought I would send this along in case anyone has encountered what I have encountered and then tried to find direction.

OS installation, GPT and MBR support

Recently, I attempted to install Windows 2012 Server to a 3Tb Intel RSTe RAID. The motherboard I was using had a C600 chipset and its BIOS supported EFI booting in addition to legacy booting.

Legacy booting implies MBR booting. This subtlety should not be overlooked.

I merrily accessed the server's BIOS and set my boot order. I wanted to be sure to boot from optical first, then the RAID. I figured I could set the optical first in the boot order, then EFI in order to get what I wanted, without any ambiguity relating to the optical drive.

F10
Save changes?
Yes, reboot.

Sure enough, I booted up to the Windows 2012 Server DVD in my optical drive, as intended, and it was off to the races. As I continued, the install process asked where to install, "the RAID!" I answered. The install process then showed me a 2Tb RAID with the balance (I had a 3Tb RAID5) unusable.

"That's weird," I thought, "must be an MBR thing. Alright Windows, do your thing and we will take it up in disk management after the fact."

No such luck, disk management would not let me convert to GPT. So, I'm throwing away storage my client had paid for.

How we lookin'? NOT GOOD.

###

Remember the statement, "Legacy booting implies MBR booting. This subtlety should not be overlooked."? What happened to me (and maybe you) is once I set the optical drive as the primary boot device (legacy) in the BIOS, my goose was cooked. The answer to my problem was to set EFI as the primary boot device, with the optical drive as the primary within EFI.

Once I made this change, the Windows installation invoked GPT support and I had the full capacity of my RAID5.

Clone and Shrink Linux CentOS 6 installation

At long last I was able to clone and shrink my Linux installation. This all came about when, originally, I created a gigantic virtual drive and later realized I did not need it all. It was excessive.

I did this on a CentOS 6 virtual machine sitting on Microsoft Hyper-V. My original VHD was 50Gb and I was using 3Gb of that. So, I created another VHD that was 20Gb. Incidentally, the original, 50Gb VHD was static and the replacement, 20Gb VHD was dynamic.

Also, in order to get started, I downloaded System Rescue CD 3.8.1 from:
Code:

and inserted it into my VM's optical drive.

Next, I attached my 20Gb dynamic VHD to IDE0 (making it sda) and my original, 50Gb static VHD to IDE1 (making it sdb).

I powered on the VM, booting to System Rescue CD 3.8.1. I had to type "startx" to get the GUI to come up and once it did, I found GPartd in the lower left. I launched it and was defaulted to the 20Gb dynamic VHD/the desired destination for this exercise. I used "Device Menu/Create Partition Table" and answered affirmatively to the prompts.

Continuing with GPartd, I accessed "Partition/New" and created a 15Gb primary partition of type "ext2". I repeated this action for a second primary partition of type "linux-swap" to finish out the drive. I right clicked the 15Gb partition, chose "Manage Flags" and ticked "boot". Then I clicked the green check at the top to commit the changes.

Also near the top, and to the right, I was able to choose the other drive- sdb. This is the source, 50Gb static. GPartd showed this drive's partitions to me, and I right clicked the large data partition I wished to shrink, then left clicked "Resize/Move" from the resulting menu. I had GPartd shrink this partition to the exact same size as I had entered for the destination previously. Again, I clicked the green check at the top to commit the changes. Then I closed GPartd.

Next, in the shell of SystemRescueCD, I entered the following commands:
Code:

cd /mnt
mkdir sda
mount /dev/sda1 sda
fsarchiver savefs /mnt/sda/source.fsa /dev/sdb1

That last command can take some time depending on the scale of what you're doing here. In any event, in my case, the command returned without errors, allowing me to proceed with my next batch of commands:
Code:

mkdir sdb
mount /dev/sdb1 sdb
mv sda/source.fsa sdb/
umount sda
fsarchiver restfs /mnt/sdb/source.fsa id=0,dest=/dev/sda1
poweroff

The gist of this is to mount the destination, output the source's partition on the destination, mount the source, move that output partition across, unmount the destination, and finally write the outputted partition to the destination.

At this point, the original, 50Gb static VHD should exist on the new, 20Gb dynamic VHD. So, the original, 50Gb (IDE1/sdb) can be detached from the VM.

Now there are some final steps, reinstalling grub and setting the swap file up in fstab. To reinstall grub, I booted to the CentOS disc/image and chose "Rescue installed system". I followed the prompts until I could get a shell and I issued:
Code:

chroot /mnt/sysimage
grub-install /dev/sda1

Now we have a booting system. At it's command prompt, I typed:
Code:

blkid

and I meticulously took down the UUID of the swap partition. I edited /etc/fstab to reflect this UUID and I was done.

Now I have successfully cloned and shrunk my install! If only I hadn't created such a monster in the first place...

ECP Issues "Sorry! We're having trouble processing your request right now"

For several days after an Exchange 2010 on Windows Server 2012 install, I ran into this error message, when accessing anything in the Exchange Control Panel (Options) via Outlook Web App:

Code:

Sorry! We're having trouble processing your request right now.

After numerous web searches, covering everything from Default Website bindings to ECP authentication schemes, I stumbled upon something odd when I opened "IIS Services Manager":

Under "Application Pools", I double clicked "MSExchangeOWAAppPool" and found a ".NET Framework Version" of ".NET Framework v2.0.50727".

I returned to "Application Pools". Next I double clicked "MSExchangeECPAppPool" and found a ".NET Framework Version" of ".NET Framework v4.0.30319".

The two did not match. Furthermore, Outlook Web App was NOT causing me problems.

After setting "MSExchangeECPAppPool" to ".NET Framework v2.0.50727" which matched the setting for "MSExchangeOWAAppPool" my problems were resolved.

Linux RAID mirror from existing

This tech article will address building a Linux RAID mirror from an existing, unraided disk. Some assumptions are made:

+There are three drives.
+The existing drive (with the original data to mirror) is /dev/sdc.
+The other two (/dev/sda and /dev/sdb) are identical make and model.
+/dev/sdb will receive the installation from /dev/sdc first. The '/' partition will be sdb3.
+/dev/sda will then receive the installation via the mirror process.

Additionally, for this exercise, CentOS 5.5 was the optical boot media. Feasibly, just about any distribution should work.

Put all three drives online, and boot to the CentOS 5.5 media. After the media has booted to the colon prompt, enter:

Code:

#linux rescue

The boot process continues. Choose the appropriate language and keyboard. There is no need to start the network, and there is no need to search for existing installations. Just choose "Skip" at the end to go straight to the rescue prompt.

Now we will duplicate the original disk's partition structure. For this example, partition 3 is '/'. But, we want all partitions anyway, so at the rescue prompt:

Code:

#sfdisk -d /dev/sdc | sfdisk /dev/sdb

This command will warn, and also output some rather dire appearing language. It's bark is worse than it's bite!

Now we want to create the RAID:

Code:

#mdadm --create /dev/md0 --level=mirror --raid-devices=2 /dev/sdb3 missing

Remember, sda and sdb will be in the RAID. sdc will NOT. So, we have just setup the RAID to include sdb. sda will come later.

Now you should be able to have a look at the RAID with the following command:

Code:

#mdadm --detail /dev/md0

The output should show one drive online/working/active (sdb) out of a total of two. The state will be 'clean, degraded'.

Now to create the filesystem on the RAID:

Code:

#mkfs.ext4 /dev/md0

This will take some processing but should finish with the output:

Code:

This filesystem will be automatically checked every 29 mounts or
180 days, whichever comes first. Use tune2fs -c or -i to override.

Now, we want to get our ducks in a row. Referencing the "assumptions" portion at the top, enter the following:

Code:

#cd /
#cd mnt
#mkdir raid1
#mkdir original
#mount -t ext4 /dev/md0 raid1
#mount -t ext4 /dev/sdc3 original

The 'raid1' mountpoint implicitly mounts sdb3, added to the RAID just prior. The 'original' mountpoint contains our precious installation. Next we will clone that installation into the degraded RAID:

Code:

#rsync -av /mnt/original/ /mnt/raid1

NOTE!!! the trailing '/' after "original". Be sure to include this or you will find yourself reworking this recipe.

The entire partition should copy from the original to the raid. When it's done we will unmount the original:

Code:

#umount /mnt/original

Now we want to add the second drive to the RAID mirror:

Code:

#mdadm /dev/md0 --add /dev/sda3

Mirroring should begin immediately. To check it's progress:

Code:

#cat /proc/mdstat

Also, be sure to edit fstab to include /dev/md0 as '/'. The original installation was not to this device.

Stay up to date

There are some indications that the phishers, hackers, black hat crowd, etc become more active during this time of year- the tax refund time of year. They would like a piece of that pie too, just like Best Buy, Walmart and Amazon.

So, they employ their usual methods, ranging from software exploits to outright (virtual) asking for your information. The latter case requires healthy doses of both paranoia and skepticism.

However, the former requires the platform a mark may be using to have all of it's updates applied. Here's the scenario:

+Gigantosoft issues a bulletin saying, essentially, "Whoops, we have a security hole in our software." These holes usually come in the form of unchecked buffers.
+The phishers, hackers, black hat crowd, etc read these bulletins. What do you think they do with the information they glean form their reading?
+Unpatched machines are exploited by malicious webpages amongst other methods.

The moral of the story? See the title of this entry- "Stay up to date".

MCS, Inc. is available to help in this regard, use the Contact page to inquire.

Information Technology

The field of Information Technology (IT) is vast. IT facilitates efficient business processes while also presenting profound challenges. Defining IT for your business can seem overwhelming.

The Information Technology Association of America has defined information technology (IT) as...

the study, design, development, application, implementation, support or management of computer-based information systems

By this definition, seven broad subcategories of IT are made apparent, each of the seven having it's own merit. In order for a business to have stable information technologies, it must integrate each of these areas to varying degrees. It is this integration that Modern Computer Service Inc. brings to the table.

When a business employs MCS, Inc. to integrate these areas, it frees resources for it's own objectives. Contact MCS, Inc. today, for an initial meeting and evaluation.

We are still here!

The earth is sill intact, there has been no Mayan doomsday. Modern Computer Service, Inc. is still operational and can prevent a cyber doomsday in your corporate IT environment.

Website Makeover

The corporate website has been reworked. The new site displays the same information, but in a better format. Gone is the php slideshow of the corporate clean theme, replaced by the more modern, views slideshow.

Other enhancements were also made and hopefully all of the kinks are worked out!

Slideshows

A plateau has been reached in MCS, Inc.'s slideshow research. Any given slideshow now has these finishing touches:

+Main slide is a weblink.
+Pager is comprised of main slide thumbnails.
+Pager thumbnails alter main slide upon hover.
+Pager thumbnails are arranged left to right instead of top to bottom.

Have a look at http://halshydraulics.moderncs.net for an example.

All of this is happening within the Drupal Content Management System. If your business has content it would like to display in this format, why not use the Contact page to learn more?

Pages

Subscribe to MCSBlog