Archive for August, 2005

HP – Part II

Well, I guess it was as simple as opening the unit’s case, flipping switch #6 to ON and rebooting. That cleaned out the CMOS and NVRAM and after that success. The machine now boots and runs SuSE 9 Enterprise. Did I say as simple as? Actually, quite a few things were tried (switching to Enterprise version, downloading updated drivers from HP.com, switching from grub to lilo, and so on. But the last thing I tried (blowing out the CMOS & NVRAM) seemed to make it reboot reliably.

I’m guessing either HP had some settings in there that worked for their default OS (Windows) but thwarted reliable Linux booting…or I changed something I shouldn’t have touched (also quite possible).

In any case, we now have our proxy server software (EZProxy) running on the new Proliant box. The old Ultra II is standing by (network cable unplugged), ready to leap back into action if we see a hiccup during the next day or so but I think it will run fine. Thankfully, since the new machine has the same name as the old, our SSL certificate works as expected. Hope I remember to update it next March.

There’s still a quirk somewhere that I’ll continue to investigate (seems it won’t boot up if we put the 2nd hard drive in the unit…or maybe it’s when the CD isn’t inserted…or maybe…well, there were just too many variables involved on the last reboot to be able to narrow it down).

With the way power comes & goes around here, we’ll have ample opportunity to test various theories during the coming weeks.

12 hours later…

Put the 2nd SCSI disk back in the server but in slot 4 (instead of slot 2). Showed up and I did a reiserfs on it…all’s well. I’m now able to say HP can make more than printers–but not yet willing to say I’d buy another one of these things. Of course, after spending so much time getting it to work, I probably should buy only these units–I now know so much about them.

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

Apple takes notice of OSX on x86 videos

The other day I referenced the French site that showed OSX running on an x86 laptop. Mes amis are back in the news today…

The OSx86 Project is reporting that Apple has served a legal notice to MacBidouille, a French news site that posted videos and instructions on running OS X on x86 hardware.

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

HP – Have Patience? Horrible Platform? Help Please?

I am about to give up on this box. I think a misplaced sense of gratitude to HP for their work developing DSpace made me do it…and now I know why you should only make rational computer decisions.

I recently ordered a new HP ML350 Proliant server (dual 3.0 Xeon processors) and a copy of Novell’s SuSE 9.3 Linux. My plan was a cool replacement for our aging Sun Ultra II (dual 300Mhz UltraSPARCs) that we use for EZProxy work. The Sun machine is so old (8+ years) that its memory chips are wearing out.

Sad to say, this HP box has been nothing but trouble. The hardware assumes you’re running Windows and some of the assumptions (right down to the BIOS) make running Linux impossible. According to a few pages I googled, it is possible to get SuSE running on this machine but you have to load all sorts of firmware updates and a variety of workaround patches from HP (many of which still carry Compaq names). HP’s website says that the system supports SuSE 9.x (both 32 bit & 64)…yeah, and my watch supports time travel.

I tried a few workarounds. For a time it seemed that the system was just getting confused by the SCSI controller, so I decided to put an IDE drive in the box, do yet another install and thereafter just boot from that. I mean, a proxy server doesn’t use the disk all that much. The install to IDE went well (I booted off the CD to get it started) but when I finished, I belatedly realized that the BIOS does not offer “boot from IDE” as an option. Arghhhhh!

I’ll try again tomorrow and if it still fails, I guess we’ll have a really top-o-the-line Windows server in the office. Perhaps we’re about to deploy a really awesome scanning station for our e-reserves folks…assuming that under Windows this thing will recognize a USB scanner.

More later…

…ok 24 hours later. Today was a bust as well. The machine will install but can not boot unless the initial install CD is present. To rub salt on these wounds, we tried a quick & dirty install 0f Windows 2003 Server…and wouldn’t you know it, it came up & ran great! Of course, it probably took on 3 or 4 worms before we shut it down but who cares, we’re reformatting the disk every 15-30 minutes.

Today I finally realized that the HP website *only* talks about SuSE Enterprise Edition (not Professional) so maybe that’s the problem. I ordered a copy of Enterprise late in the day and when that comes in I’ll give it the final attempt.

Given this blog’s readership I don’t expect an outcome much better than if I put a message in a bottle and set it adrift…but if google caused you stumble across this and you know *anything* at all about SuSE Linux and Proliant ML350 G4 (not the smart array model) servers, please leave a comment…

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

Free e-books

Lots of O’Reilly titles for free web-based viewing…also a few WORX titles and others. Great site:

ebook 911

http://ebook911.com

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

Wotsit’s Format?

wotsit's format?

Just discovered Wotsit’s Format…a website devoted to information on many file formats. From the introduction:

Welcome to Wotsit’s Format, the complete programmer’s resource on the net. This site contains file format information on hundreds of different file types and all sorts of other useful programming information; algorithms, source code, specifications, etc.

Assuming they know more about file formats than grammar, this is just the sort of site we’ll need to keep track of for future file migration strategies in MARS and follow-on systems. It seems to me some library group should be focusing on this (e.g., set up a clearinghouse operation that gathered information on every known file format in existence). Perhaps some library entity is doing this and I just missed it. I do know the Long Now Foundation is beginning to work on a Wikipedia-like system for this information. A factoid I uncovered in this interesting thread that appeared on the Omidyar Network.

To get an idea of why this is already a problem, take a look at this article “The Fading Memory of the State” by David Talbot (from MIT’s Technology Review).

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

Mason makes IT news again

hey...my big brother goes to that school...

OK, I guess this is an improvement over the last time Mason made IT news

From C/Net News

The federal government is funding the development of a prototype surveillance tool by George Mason University researchers who have discovered a novel way to trace Internet phone conversations.

Their project is designed to let police identify whether suspects under surveillance have been communicating through voice over Internet Protocol (VoIP)--information that would be unavailable today if people choose to communicate surreptitiously. The eavesdropping technique already has been shown to work with Skype, the researchers say.

"From a privacy advocate's point of view, this is an attack on privacy," Xinyuan Wang, an assistant professor of software engineering and principal investigator, said Tuesday. "From a police point of view, this is a way to trace things."

To translate his research into a tool that could be used by police in a successor version of the FBI's Carnivore system, Wang received a grant of $307,436 from the National Science Foundation this month. The grant calls for the development of a prototype VoIP-tracing application to provide a "critical but currently missing capability in the fight on crime and terrorism."

The NSF grant comes as federal police are fretting about criminals using VoIP to mask their communications. The Federal Communications Commission on Friday approved mandatory wiretapping requirements for some VoIP providers, and the FBI has been warning for more than two years that VoIP may become a "haven for criminals, terrorists and spies."

At the moment, two Skype users who wish to conceal the fact that they're chatting can direct their computers to bounce their conversation off a commercial anonymizing service, sometimes called a proxy service. Such services are offered by FindNot.com, Proxify.us and Anonymizer.com.

The FBI or any other government agency that's eavesdropping on both ends of the link would see that each person was connected to the anonymizing server--but couldn't know for sure who was talking to whom. The more customers who use the service at once, the more difficult it would be for investigators to connect the dots.

Wang discovered he could embed a unique, undetectable signature in Skype packets and then identify that signature when they reached their destination. The technique works in much the same way as a radioactive marker that a patient swallows, permitting doctors to monitor its progress through the digestive system.

"It's based on the flow itself," Wang said. "I embed a watermark into the flow itself, the timing of the packets. By adjusting the timing of select packets slightly, it's transparent. There's no overhead in the bandwidth, and it's very subtle. It's mingled with the background noise." (The anonymizing service tested was Findnot.com, which did not immediately respond to a request for comment on Tuesday.)

A paper co-authored by Wang and fellow George Mason researchers Shiping Chen and Sushil Jajodia describing their results is scheduled to be presented at a computer security conference in November. An early draft concludes that "tracking anonymous, peer-to-peer VoIP calls on the Internet is feasible" with only 3-millisecond timing alterations as long as the calls are at least 90 seconds long.

Peter Wayner, an author of books on cryptography who is teaching at Dartmouth College, predicts that an arms race could develop between VoIP programmers and eavesdroppers. The George Mason research "seems as likely to yield new techniques in anonymizing as it is to yield techniques for stripping people of their privacy," Wayner said.

"I think it's pretty academic right now," Wayner said. "It would take a lot of work to track people. They'd have to really be interested in someone to use it."

The George Mason researchers' technique does not try to decipher the contents of encrypted conversations (Skype, VoicePulse and PeerMe are VoIP providers that use encryption). In other words, it tries to glean only the identities of the participants, not what they say.

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

Clone a Solaris boot disk

Having a disk failure the other day reminded me I need to create a clone of the boot drive in our V880 machine…so recovery from the inevitable disk failure won’t require a full reinstall of Solaris (had to do that a year or so ago and it took way longer than I wanted it to). Since I’m not a fan of tape backups, I want to have a boot-drive cloning system in place (like I use for Linux & Mac OSX Server machines). Haven’t found a Carbon Copy Cloner for Solaris yet, so here’s the next best thing.

Here are my notes thus far:

1) Install a duplicate of the boot drive … same geometry, and so on. This part isn’t too hard in the standardized Solaris hardware realm. This isn’t critical but it makes things a lot easier. If you don’t have an identical drive, you can’t use the prtvtoc command shown below.

2) Partition the new drive exactly like the existing one. You can do this via the format command…counting cylinders and so on…but here’s a much faster (and less error-prone) way to do that:

Say the original drive is c1t0d0
and the new drive is c1t4d0

prtvtoc /dev/rdsk/c1t0d0s2 | fmthard -s – /dev/rdsk/c1t4d0s2

Voila…new drive is partitoned exactly like the orginal.

Now, this script will build new file systems on the partitions of the new disk…matching those on the original. Then it will run ufsdump, copying data from the original to the new drive…then unmount the new “clone” drive. Finally it makes the new clone bootable. Note that this jazzy blog format wraps lines on the installboot line of the script…a “man installboot” will give you a clean copy of the syntax.

#! /bin/ksh
# script assumes:
# c1t0d0 is original
# c1t4d0 is drive we’ll turn into a clone

partlist=$(prtvtoc /dev/rdsk/c1t4d0s2 | awk ‘!/\*/ {print $1}’)

for p in $partlist
do
if [ "$p" != "1" -a "$p" != "2" ]
then
newfs /dev/rdsk/c1t4d0s$p < /dev/null
mount /dev/dsk/c1t4d0s$p /mnt
cd /mnt
ufsdump 0uf - /dev/dsk/c1t0d0s$p | ufsrestore rf -
cd /
umount /mnt
fi
done

mount /dev/dsk/c1t4d0s0 /mnt
installboot /usr/platform/`uname -i`/lib/fs/ufs/bootblk \
/dev/rdsk/c1t4d0s0
umount /mnt

exit 0

Update (9/23/2005): The other day our power went off (again) so I decided to test whether the clone drive I created using this process actually worked as I had hoped. Pulled the disk out & put it in the boot drive’s slot (0) on the V880…when power came back on I hit the switch…yes…it booted fine and loaded Oracle, Voyager and everything else without a hitch.

Shut things back down & swapped the drive back out of the boot position–but now I can sleep a bit more peacefully.

Add to Del.icio.us Add to Technorati Stumble Upon Digg This

Next Page »