Archive for the ‘hardware’ Category

Fast, or Floppy?

October 16, 2019

In 1999, I had a DEC Celebris mini-tower for my desktop PC at UUNET. It had a Pentium Pro 200 MHz CPU, which was top of the line when the PC was purchased in 1996. But after several years, it had been far surpassed by newer PCs. PC hardware was advancing rapidly and the useful life of a PC was 3-4 years before it was obsolete.

catapult DEC PC under Dave's desk

I was managing a group of web hosting developers, but I didn’t have the budget for buying new desktop PCs. My biggest problem with my PC was that it was too slow to play MP3 files while doing anything else, and those were all the rage then. One of my co-workers had discovered Napster.

Then I learned about a CPU upgrade that adapted a Pentium II-based Celeron CPU to a Pentium Pro socket: a PowerLeap PL-PRO/II at 667 MHz, more than doubling the CPU power. I ordered a couple. A week later, they arrived at the office.

The Celebris parts weren’t fully interchangeable with ATX PCs. The CPU was on a horizontal daughterboard and the screws and plugs and I/O plates were all a little different from the standard. But the upgrade CPU fit. Except… it had a big heat sink and cooling fan that needed to be right where the floppy drive was on the Celebris. On a standard ATX PC the CPU wouldn’t have had a clearance problem.

This was before USB and CD burners became widespread, so we actually used the floppy drives. I had a choice to make: fast, or floppy? I removed the floppy drive. I didn’t have a blank cover for the hole left by the floppy drive, so it became a large ugly air intake. That’s a price I was willing to pay. My PC was useful again! And I didn’t have to reinstall or debug any software! If I ever needed the floppy drive to reinstall an OS, I could open the case door and hang it off the ribbon cable temporarily.

I installed the other upgrade CPU in a rack-mounted Celebris that needed more speed. In the next fiscal year, we bought some new Pentium III PCs, and we got spare Sun workstations running Solaris to augment them, but my upgraded DEC Celebris got me through a tight budget with only a little hassle. I’ve always hated to waste things that could still be useful. More than a decade later, I threw out my floppy disks and removed the last floppy drive from one of my PCs.


From Novice to Master, and Back Again

January 14, 2013

In 1985, I was a freshman at St. Olaf College in Minnesota. The college had a VAX 11/780 running 4.2BSD and a PDP-11/70 running v7 with some Berkeley and local code hacked in. It was my first experience with multi-user systems other than dialing into an MS-DOS BBS or two.

The college’s Academic Computing Center had printouts of the 4.2BSD manuals, plus some home-grown documentation, available for sale so students could learn how to use UNIX. One week I sat in the Science Center terminal room and started going through the alphabetical list of the commands available on the VAX, trying each one and reading its man page to learn what it did.

Eventually I got to “su”. “Become the super-user”? What’s that? Does it involve wearing a cape? Sounds interesting, so I tried it. To my disappointment, it just asked for a password, and wouldn’t do anything.

Shortly thereafter, someone came running into the room and asked, “Are you David MacKenzie? Did you just run ‘su’?” “Yeah… what does it do?” “Uh, don’t do that.” My failed “su” attempt had been logged on the system console and one of the sysadmins was worried about an attempted breakin.

Within a year, I did have root access on the VAX, as I learned enough to be hired as a student system programmer. I contributed to upgrading the machine to 4.3BSD when that was released.

Recently I was working on a CentOS Linux virtual machine and needed to look up the command-line options to “su”. I had worked for the past several years mostly on Macs where “sudo” is preferred, so my “su” skills were rusty. I ran “man su” and got the information I needed. Then at the bottom of the screen I sheepishly read “Written by David MacKenzie.”

In the 1990s, while filling in gaps in the GNU toolset, I wrote the GNU “su”, and I had forgotten about it. It’s still what Red Hat and other distributions are shipping.

At least I know what it does now.

Protecting Cable Service from Surges

February 9, 2012

I found out a few months ago that cable TV/Internet service is supposed to be grounded at the entrance to the building. I learned the hard way, when during two storms, the Comcast Business cable at my office was hit by surges (I suppose from lightning strikes somewhere exposed upstream). The surges killed our cable modem, which Comcast had to send a technician out to replace and reconfigure each time. That part cost us a day or so of downtime in each storm. Worse, the surges traveled down the Ethernet cable from the modem and destroyed other equipment. The first storm fried one Ethernet port in our firewall appliance, and the second strike killed all the Ethernet ports in our firewall appliance and a packet shaper rate limiter attached to it, equipment worth around $8,000 which we had to replace and make an insurance claim for.

After that, I did some research about what could be done. It appears that the National Electrical Code requires communications cable lines to be grounded at the building service entrance, as referenced in this Q&A from the New York State government and this Q&A from the Electrical Contractor Network. I am not an expert in the topic.

I came up with a two-pronged defense. The first was, we had Comcast send another technician out to ground the cable to an electrical panel where it enters our server room. He affirmed what I had read on the net, that the cable is supposed to be grounded at the building service entrance. He said that Comcast used to do it by driving a rod in the earth, but that now grounding to an electrical panel is preferred. He demonstrated the constant voltage differential between the two by touching the grounding wire against the splitter lug while the other end was screwed to the breaker panel; we saw little sparks arc.

The second prong of defense was to install an Ethernet surge protector on the CAT5 run between the cable modem and the new firewall, so even if a surge did hit the cable modem, it wouldn’t continue up the line to damage other equipment. We got the APC PNET1GB ProtectNet Standalone Surge Protector for 10/100/1000 Base – T Ethernet Lines for around $24 from Amazon. It uses a grounding wire run to an electrical outlet’s screw.

After getting the office taken care of, I implemented the same measures at home. We happen to have a Comcast-provided splitter where the cable enters our house, with a grounding lug on it, and a corner clamp for grounding already installed on the electrical meter box just above it. So all I needed to do was run a copper wire between them. I also got the Ethernet surge protector noted above to run between our cable modem and firewall/router.

So far, so good; no damage during storms at either place. It’s sad that Comcast doesn’t consistently protect customer lines when they do an install, though. If you have a cable line that isn’t grounded, consider calling your cable provider and asking them to send someone out to do it, and/or install an Ethernet surge protector yourself.

Here are some pictures of the installation at home.

Grounding cable run

Grounding clamp on electrical box

Grounding lug on cable splitter