Network Interface Aliases in Solaris

Most of what you will find when searching the net for “solaris network interface alias” or “solaris interface alias” will be ways to create pseudo interfaces such as “hme0:1” or “ce0:1”. This is a well documented feature under Solaris.

However, in my case, we were looking at ways to incorporate other systems in to our RAC cluster. Due to the way that Oracle RAC is designed, the public (primary) interface on all of the systems in the RAC cluster MUST be the same. In other words, they must all be “ce0” or “hme0” or whatever the device name is.

This isn’t really a problem under linux since you can easily define which interface gets to be named “eth0”, but under Solaris, this can be a bit of a problem when you try to incorporate different machine types in to the cluster. Because Solaris assigns an interface name based upon the chipset of the NIC, some systems might have “ce0” (such as on the V480) while others might have “ipge0” (such as on the T2000).

So.. how do you add a machine with a different primary device name in to the cluster? Well, the answer is actually quite simple: create an alias to it. It’s just not the same alias that you get from plumb’ing an interface.

In the /dev tree, simply create a symbolic to the primary interface using the name that you want. For example:

# cd /dev

# ln -s ce public

# ifconfig public0 plumb

# ifconfig public0
public0: flags=1000842<UP,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
inet 0.0.0.0 netmask 0
ether 0:3:ba:2a:e8:c5

And you are done. You now have an interface called “public0” which is known as your primary interface. Be aware, though, that this will BE DESTROYED if you reboot the system with the “-r” flag as this will force a rebuild of the entire device tree.

Replacing a Failed Disk on a Network Appliance (Netapp)

Replacing a failed disk on any system can be a scary thought, but a lot of technology has been put together on the NetApp to make this as easy and painless as possible.
Without going in to all of the talk about what RAID levels to use and the use of DP (dual parity) disks, there is an easy procedure for replacement. If the disk has not completely failed, you may see messages like:

Disk Inventory Monitor: Dual Loop Configuration WARNING!!!
The system has detected an inconsistency in the device maps
of paired-channels listed below. Review the device maps of
both channels to check for disks which may not be responding
on both A and B device ports.
channels: 0c and 0a.

In this case, you will want to find out what device is failing. Execute :

storage show disk -p

The output will look like:

PRIMARY PORT  SECONDARY PORT SHELF BAY
------- ----  --------- ---- ---------
0a.16    A    0c.16      B     1    0
0c.17    B    0a.17      A     1    1
0c.18    B    0a.18      A     1    2
0c.19    B    0a.19      A     1    3
0a.20    A    0c.20      B     1    4
0a.21    A    0c.21      B     1    5
0c.22    B    0a.22      A     1    6
0c.23    B    0a.23      A     1    7
0c.24    B    0a.24      A     1    8
0c.25    B    0a.25      A     1    9
0c.26    B    0a.26      A     1   10
0c.27    B    0a.27      A     1   11
0c.28    B    0a.28      A     1   12
0c.29    B    0a.29      A     1   13
0c.32    B                     2    0
0c.33    B    0a.33      A     2    1
0a.34    A    0c.34      B     2    2
0a.35    A    0c.35      B     2    3
0a.36    A    0c.36      B     2    4
0a.37    A    0c.37      B     2    5
0a.38    A    0c.38      B     2    6
0a.39    A    0c.39      B     2    7
0a.40    A    0c.40      B     2    8
0a.41    A    0c.41      B     2    9
0a.42    A    0c.42      B     2   10
0a.43    A    0c.43      B     2   11
0c.44    B    0a.44      A     2   12
0c.45    B    0a.45      A     2   13
0d.16    B    0b.16      A     1    0
0b.17    A    0d.17      B     1    1
0d.18    B    0b.18      A     1    2
0b.19    A    0d.19      B     1    3
0b.20    A    0d.20      B     1    4
0b.21    A    0d.21      B     1    5
0d.22    B    0b.22      A     1    6
0d.23    B    0b.23      A     1    7
0b.24    A    0d.24      B     1    8
0d.25    B    0b.25      A     1    9
0b.26    A    0d.26      B     1   10
0d.27    B    0b.27      A     1   11
0d.28    B    0b.28      A     1   12
0d.29    B    0b.29      A     1   13

Notice that disk 0c.32 is only showing up on one channel. That is the one that is causing the problem. Let’s fail the disk over to the spare:

disk replace start 0c.32 0d.29

The system will prompt you with the normal “are you sure” message, and, with the positive response, the system will begin the process of replacing the failing drive and rebuilding the contents on the new drive.

You can monitor the progress with “sysconfig -r”.

Once the reconstruction has completed, you will want to remove the drive. In order to help you properly identify the drive, you can have the RED LED blink on the drive in a consistent manner to make it obvious to the person who will be pulling the drive.

priv set advanced
blink_on 0c.32

priv set admin

If the disk is completely broken and the LED won’t blink, just turn on the RED LED’s on either side of the bad disk. That way, the one in the middle (not blinking) is easy to find and replace.

Now that we are ready to replace the drive, execute:

disk swap

Pull the drive from the system and insert the new drive.

disk swap

And your new drive has been added to the system as a spare. Confirm this with ‘volstatus -r’ or ‘sysconfig -r’.

MythTV: The Quest For a Quiet PC

MythTV, if you follow the design principle, is best constructed with multiple machines. This separates your BackEnd (BE) server process from your FrontEnd (FE) server process and allows for the entire setup to be completely client/server based over a network.

Not sure of what I mean? Well, the idea is relatively simple:

  1. Set up one big, fat, noisy machine as your BackEnd. Allow for plenty of expandibility because, in the future, who know what you might want to have MythTV do. Also, because it is a BackEnd (BE), you don’t care too much about how it looks or how noisy the drives and fans are because this system will be stuffed in a corner or in a closet for no one to see or hear. (Be sure it is a place with adequate cooling, though!)
  2. Set up any number of sleek, pretty, and quiet machines throughout your house (with network access, of course) as you FrontEnd (FE) systems connected to some displays.

Now, nearly everyone who is playing around with their own Tivo-like homebrew setup is probably very familiar with the first machine (the BE). But, the other machine(s) (the FE) are a whole different ball game. Simple things can get expensive here, and the more silent you wish to have your FE, the more expensive it will get. People can go nuts over this (take a look at SilentPCReview.com), and for good reason: you don’t want a noisy machine ruining your viewing experience.

The real question that you have to ask yourself is: How quiet is quiet enough? And remember, Quiet is not equal to Silent. Silent is equal to very expensive. Quiet is equal to not as expensive. I can only go by my tastes, but I tried out a couple of different cases before I got to one that I felt was acceptable.

overtureii.jpgFirst, I bought the Antec Overture II case. It’s a nice looking case with a sharp Piano Black finish and is very quiet. However, right after I got it, I realized that it is just plain HUGE. It’s like taking a full-sized ATX case and laying it on its side. The airflow in the unit is awesome and it has bays for everything. Unfortunately, that is not what I wanted. I wanted a small system that would be, well, for all intents and purposes, nearly bare-bones. I mean, everything was going to be on the server, so I didn’t need something capable of holding terabytes of data at the frontend.

Ultra MicroFlyAfter selling that on eBay, I bought a smaller case off of TigerDirect.com. I picked up
an Ultra MicroFly case with a supposedly quiet 400Watt power supply. Now, when it comes to bargain stuff, I usually take the words “quiet” and “near silent” to be worthless marketing buzzword bingo. And, yes, this goes for the MicroFly. “Quiet” is a relative term, and, in MicroFly, “Quiet” can be defined as: Being less noisy than a tower system.

shuttle_sk21g.jpgBeing dissatisfied with this purchase, I sold it on ebay and tried out a Shuttle SK21G from Newegg.com. No, I wasn’t all that happy that the case was silver. I’m a “black-entertainment-center-equipment” type of person, but, I was somewhat tied to the AMD64 754 socket since I had two processors of that design laying around. I also wasn’t all that thrilled with getting a custom case with a custom motherboard, but I figured that if I didn’t like it, I could re-sell it on ebay, too.

However, after the quick assembly, the SilenX + ICE setup that the Shuttle has, the system is very quiet and cool.  The assembly process really isn’t much since the motherboard is already in place, but you still have to be careful and take your time.  I realized that the cabling was already run for 1 SATA device (a hard drive) and 1 IDE device (an optical drive).  Since I only had IDE drives, I was forced to use the IDE cable and not install an optical drive.  I figure that I won’t be using the system as a DVD player just yet anyhow.

With this case, there is no processor fan.   This case comes with what they call SilentX and ICE, which is really a water cooled system for the processor.  With that in mind, you have to figure that will impact the price of the system, but, when you compare it with comparable systems, you’re really not paying all that much more.  Because of the water cooled system, there is a large radiator at the back of the case (inside) which has a smart (variable speed) fan strategically placed to cool it off.

My main fear with a Shuttle system was that they would run much hotter than a normal system.  I quickly discovered that this was not the case since the processor ran around 37°C and the case fan speed hung around 950RPM.

The system’s physical size and overall looks made it ideal for my purposes.  It fits right in to most entertainment centers and can be easily hidden away.  It doesn’t pull much power, especially since I am using AMD’s Cool-n-Quiet technology with Fedora Core 64, which happily slows the clock speed down and speeds it back up on demand.  This translates in to the system running cooler, so the case fan runs slower, and, hence, quieter.  That coupled with the liquid cooling really makes this system comfortably quiet.

I’m satisfied with this case as a good choice for me.  It wasn’t until after I set the case up that I discovered that Shuttle has an ebay store where they sell refurbished units.  The cost of a refurbished unit was a lot lower than NewEgg‘s price on a new unit, and, since I don’t mind having a factory refurbished system, I bought my other two cases from them.

Keep in mind that the noise level of any given case is also directly dependent upon the other items in the case: hard drive, optical drive (not so much since it isn’t constantly running), and video card.  Because of the application I am using these systems for, I had to use nVidia cards, and I noticed the sound difference between my PNY 6200 with the bun-fan on it versus the fanless XFX 6200 I have.  I also noticed an immediate difference between an older hard drive that I had versus a new Hitachi DeskStar drive I just picked up (the Hitachi is dead quiet).