Many things happening @ Cray-Cyber.org, most of it behind the scenes, though. We found some more additional storage space for the collection in second half of 2008 and already started to fill it up. Currently we have several SGIs in there (mostly O2000 and Onyx2) and lots of "new" CDC spare parts and documentation. Right at the moment, John is out of Germany with a team of people to unmount another huge Control Data installation. Parts will end up in the new storage facility on Wednesday of upcoming week. Maybe pictures will follow, stay tuned.

Upcoming actions for our small team in the computer room: fresh install of our Cray YMP-EL, which currently has a broken system disk, as you might have noticed already. Fortunately, we have enough spare disks here. Another main task will be the setup of our Cray J916 and finally fix another broken system disk in our SGI Origin 2000. Since it's new year, we of course have plenty of New Year's pledges :-)

Oh, and for the Web 2.0 freaks out there: We recently set up a twitter stream. Feel free to follow us at http://twitter.com/craycyber.

Yesterday even before leaving home for operating, I experienced a non working web server and no logins possible at Cray-Cyber.org. So, first thing right after arriving at the site was to check our Xen server. Everything looked fine except the webserver virtual instance. Tried a Xen reboot of that instance, but no success. Finally I rebooted the entire server with all Xen instances. Everything came up fine, all systems running. One thing I couldn't get resolved though, was DHCP over WiFi. Instead of debugging for hours, I just took a nearby ethernet cable an will do debugging another time.

Suddenly during afternoon, yel came up with a temperature error. When looking on the machine, all I got was a telnet connection refused error. So, let's check the console. Uh oh! Told me something about unrecoverable read error on disk. The operating system couldn't load /etc/inittab it said. Here you can find a screenshot of the errors. I was hesitating to switch off the machine and reboot. Mostly, the root cause for errors like this at such vintage machines isn't obvious and in most cases some really broken hardware. When Alex arrived, we decided to reset the CPU and reboot UNICOS. The Cray booted up without noticable errors, is now up and running again and happy to service you.

Last week, one of two control boards in our E10k refused to work. The redundant one spontaneously decided to do the same. No control board, no running E10k. So today I investigated.

First: switch off power on everything in the E10k chassis. Connect to all important log files to see what'll happen. And power on control board 0. Decides to boot this time. Cool. Next switch on power to our private domain, consisting of only a single board (4 CPUs). Bringup the domain. Everything fine, Solaris boots, login available. Fine.

Well. Then there is another apparently broken control board 1. Switched on that as well. Boots just fine. Now I'm getting courageous. There is our public domain, consisting of 56 CPUs, distributed among 14 system boards. Switched on power to all of them and entered the bringup command. Whoow! Works! Great! I troubleshooted and repaired two control boards the same time by doing plain nothing! Cool thing :-)

A good engineer will make some quality assurance to his work at the very end. So I decided to switch of the entire platform and start over. This time according to our self written boot manual. And well, everything works as expected. Both domains are up and running, accepting logins, have all their filesystems mounted and so on.

To make a long story short: We suspected a broken E10k last week, but couldn't find any errors or other hints. Everything is up and running again. Lucky we.

we are proud to announce that we finally received a replacement Ethernet-Board for our Cray Y-MP EL (if someone of you happens to have another one laying around, please tell us! We urgently need spares).
Because of this we were able to take the Cray Y-MP EL back into operation. From today on this machine will be running again 24x7. All user-accounts as well as our guest account are also available on the Cray. To use the Y-MP EL, choose number 1 (hostname yel) from the Host-Chooser-Menu and log on using your username and password. Regarding the performance, please remember that this machine has Vector-CPUs. For more information, please see: http://www.cray-cyber.org/systems/yel.php)

Furthermore majestix, our Control Data 4680 is also back in operation every Saturday. Due to a hard disk failure we needed to reinstall EP/IX 2.2.1AA and recompile and port all the layered products. Thanks to Gerard van der Grinten who spent one week here at Cray-Cyber.org and reinstalled majestix and fixed some of the tape drives of the Cyber. More information about the Control Data 4680: http://www.cray-cyber.org/systems/majestix.php

That means that we now have the following machines in operation:

• Cray Y-MP EL running UNICOS 9 (yel) 24x7
• Cyber 960-32 running NOS and NOS/VE (cy960) - every Saturday
• Sun Enterprise 10000 running Solaris 10 (e10k) - every Saturday
• Control Data 4680 running EP/IX 2.2.1AA - every Saturday.

We are still working on taking the SGI Origin 2000 back into operation and will announce its availability as soon as we are ready.

Have fun using our systems and don't hesitate to ask us if you have any additional questions. Best Regards,
Your Cray-Cyber-Team

A vast amount of our documentation is currently located on microfishes. We of course have microfiche reader, so access to this documentation isn't a problem for us. But! Wouldn't it be much more convenient, to have access to all this information electronically? With the PDF viewer of choice? Maybe even with full text retrieval? Well... of course, this would be great! So, we did a first effort this weekend on the way. We startet investigation on a high quality easy to use way of digitizing microfiches. First idea was, to just take high resolution pictures directly of the microfiche reader screen. To surpress interference of the surrounding light, we also did some tests in the dark. Both test series showed useful results, although quite a lot of post processing will be neccessary. We will take another try with a digital SLR, where we can fix focus, aperture and exposure time. This should give nearly the same light and color on each photograph, which makes automated post processing much easier.

Another try was, to use the macro mode of a digicam and backlight the microfiche. So, we created a white drawing with GIMP, turned up brightness of the laptop screen to maximum and put on one fiche on the laptop display. This proofed to not be usable. At least not on a 1024x768 resolution. Besides using a SLR, we'll also trying a modern high resolution scanner (9600x4800 dpi) with backlight addon for scanning slides.