During my evening commute, my deer smashed Monte Carlo turned 100,000 miles old.
The sad part of this story is that I went to great effort to use my digital camera to make a video of the odometer rolling over, but something went horribly wrong. Now, admittedly, it would seem that watching a digital odometer "roll over" doesn't sound very exciting, but the video turned out pretty good... right up to the point that it stopped: ten seconds too soon! I guess videos had a 2 minute limit.
Monday, November 26, 2007
2000 Monte Carlo Turns 100,000
Tuesday, November 20, 2007
One Less Deer In Maryland
...At least that's my assessment. I hit her getting on to Highway 29 at about 0600 on my way to work. I came around the cloverleaf at between 40 and 50, was speeding up to merge into traffic, and checking my mirrors. By the time I looked forward, there was no time to react.
I don't know how she made it across the three lanes of traffic or whether she made it off the on-ramp. Just not the kind of thing you think about dealing with on a DC commute.
It's hard to see in the picture, but the right headlight unit (lights, turn signal, running lights) is shattered, hood is dented, and the nose cap has several cracks.
Friday, October 26, 2007
Beowulf: Everybody Needs One
I set aside three matched systems to act as nodes. A fourth system is configured as the head. The cluster will be utilizing my standalone MySQL server and file server. I thought (incorrectly) that the head node would only be used as a control point, so it is my application server. This machine runs NIS, Bind, Apache, and Squid.
Starting point was Beowulf HowTo. It's rather old and out of date, but close enough to get things started. The single biggest factor is understanding that their is no beowulf-*.rpm it's lam-*.rpm. (Get it: wolf and lamb.) I didn't need to concern myself with NFS or NIS, as I already had those running.
Build a user, gen and distribute SSH keys. Here's a trick for that one. Since the user is going to be an NIS user, we put their home directory on the NFS server, and configure the nodes to auto mount the home directory.
# ls -s /home/cluso /net/scully/home/cluso(Yes, I know I spelled Clouseau wrong.)
# chown cluso.cluso /home/cluso
Now, gen the keys, and assign the users public key to the authorized keys file.
$ ssh-keygen -t rsaSince our own key is trusted, and we get the same directory no matter where we login, our keys always allow us in.
$ cp id_rsa.pub authorized_keys
I ignored all the MPI stuff, since I'm not doing C++ parallel computing. My goal is to be able to schedule Perl scripts across several machines. The benefit of using Beowulf for this rather than a bunch of crontabs, is that Beowulf will load balance the executions. Thus, the fastest node will be asked to execute the greatest number of incarnations of the script.
For the node configuration, two items of importance. First, disable iptables! Swendson emphasizes this fact, and yes you got to do it, or waste allot of time on rules. Besides, we're in a trusted environment. Second, the lamhosts file has moved. It is now /etc/lam/lamhosts.
Initialize the cluster with:
lamboot -v lamhostsRead the output. It took several tries to get all the nodes to respond. Since we are not doing MPI I could not use mpirun. Instead, the command is:
$ lamexec hostname
baltar.terran.lan
c17.terran.lan
c18.terran.lan
c19.terran.lan
Yeah! Oh wait... It ran on the head node also. Not what I wanted.
$ lamexec n1-3 hostnameMuch better. One other small problem is an annoying message about MPI. Let's get rid of that.
c17.terran.lan
c18.terran.lan
c19.terran.lan
$ lamexec n1-3 hostname 2> /dev/null
Now for the real fun. Lets execute 10 runs, without concern for the target nodes.
]$ lamexec -np 10 hostname 2> /dev/null
baltar.terran.lan
c17.terran.lan
c18.terran.lan
c19.terran.lan
baltar.terran.lan
c17.terran.lan
c18.terran.lan
baltar.terran.lan
c19.terran.lan
c17.terran.lan
since we can specify nodes as either n1-3 or n1,3, we have significant control over the processing capacity.
Man, that was easy.
Monday, October 15, 2007
NH Rest Stop
This has got to be my favorite rest stop of all time. Above this Rest Area sign on I-93, is a small sign mentioning a state run liquor store is at this rest stop. Yes, a liquor store, at the rest stop, run by the state. Brilliant!
View Larger Map
What is truly amazing, however, were the prices-- 33% less than I pay in Maryland. Too bad I'm flying home.
Tuesday, October 09, 2007
"Do Not Push"
Saturday, October 06, 2007
Xen and Fedora 7, Pt 2
The problem is one of memory allocation. Under Red Hat Enterprise Linux 5, you must specify more than 512M for an installation. Anything less results in a "not supported" error. On Fedora 7, virt-manager defaults to 500M. If you choose a lesser value, you are threading on thin ice.
There seems to be a sweet spot between 396M and 431M. With memory set to less than 448M, the installation warns that there is not much physical memory, and asks if it can eneable the swap partition for use by the installer. At 256M the system will freeze and the install fails. At 431M, the installation proceeds. Several attempts as 293M resulted in about 50% freezes.
Sorry to say I didn't bang away at it long enough to isolate the exact break point, but there is really no need. If you are critical on memory, and must do an interactive install (as opposed to kickstart), install at 448M, and bump down for production.
Xen and Fedora 7
[root@adama ~]# virt-install -n apollo02
-f /dev/vg0/xen-apollo01 --nographics -p
-l ftp://192.168.69.13/f7-64 -r 520
-x "ks=ftp://192.168.69.13/ks/apollo.ks"
Starting install...
... output truncated ...
Write protecting the kernel read-only data: 938k
And the guest locks. Or does it? In reality, the guest is running the kickstart install but the console can not connect. If we try xm console X, we get nothing. If we try to connect through virt-manager, we get nothing. But, as long as the kickstart is fully automated, the install will work.
Let's try this:
[root@adama ~]# virt-install -n fc6vm
-f /dev/vg0/xen-apollo01 --nographics -p
-l ftp://192.168.69.13/fc6-64 -r 520
-x "ks=ftp://192.168.69.13/ks/fc6vm.ks"
Starting install...
... output truncated ..
Welcome to Fedora Core
The console launches, and everything works as expected.
Interesting. So... Whats different between the was F7 boots and FC-6? One thing I had noticed was that F-7 boots in the XGA mode (43x80 vs 25x80). I try adding text to the -x option, without luck. Then I tried this:
[root@adama ~]# virt-install -n apollo02
-f /dev/vg0/xen-apollo01 --nographics -p
-l ftp://192.168.69.13/f7-64 -r 520
-x "ks=ftp://192.168.69.13/ks/apollo.ks console=vga"
Starting install...
... output truncated ..
Welcome to Fedora
Much to my elation, it worked. One more thing to check, however.
Stay tuned.
Casarsa Chardonnay
Smooth as can be. This Italian chardonnay is value prices, packed in 1.5 liter bottles and oh so sippable. 7 of 10
Friday, September 28, 2007
Fedora 7 Services
services disable anacron atd avahi-daemon bluetooth cpuspeed dhcdbd firstboot hidd hplip iptables ip6tables nfslock pcscd rpcbind rpcgssd rpcidmapd
Wow! That's allot of bloat.
Thursday, September 27, 2007
Xen: Fedora 7
I did two things different:
1. Allocated 520M of memory for the install. Fedora Core 6 would install with 256M, but 7 would fail.
2. I selected the default partitioning scheme. Unfortunately, the default scheme was thought up by a brain dead engineer. My evidence is that they set aside all disk space for LVM and allocate it all to root. Turns out, root is the only file system you can not reduce. Thus, no way to reorder the partitioning.
I think it was number 2 that made the difference, as Disk Druid would run, ask to activate swap, than freeze regardless of whether I picked Yes or No. Should be easy enough to prove, but that will have to wait for tomorrow.