Cisco shocker: Some network switches may ELECTROCUTE you Oh dear: Cisco is warning that screws in a couple of its compact Catalyst switches may be poking into wires carrying live voltages. In this field note, the Borg says the problem occurs when WS-C3560CX or WS-C2960CX switches are installed without a mounting tray – for example, screwed to a desk, shelf, or wall. Screws not …
If I were designing a case I'd make sure that no screws that a user can insert from the outside are co-linear with a mains voltage source. Assume that even if the supplied screws are 5mm long, somebody will try to use 50mm screws, and that removing him from the gene pool won't be good for corporate profits even if it does improve the average IQ of the human race.
OTOM, making sure that one of the screwholes is co-linear with the master CPU or switch chip might work wonders for profit margins ....
> The installation instructions are quite clear ...
Then the design is FAIL. Even if the first user reads the instructions on opening the box, there is no way that they won't get separated from the unit fairly shortly afterwards - lets face it, most of us are used to keeping anything on the shelf that finds itself free and using it when the need arises.
I can't say I've ever taken something off the shelf and thought ... "Hmm, must search for the instructions in case some designer fooked it up and made it dangerous if I don't use specific screws". No, if it needs mounting, I'll look at the mountings and see what I've got that'll fit - if the design doesn't cope with that safely then the designer is 100% to blame. It's not like this is rocket science.
Thinking back to something less dangerous, I recall many many years ago when we got a new DAT drive for the unix box - back when 2G DAT was "huge". Some fecking idiot put a ribbon cable right behind a mounting screw hole and they "covered their arses" by specifying the maximum screw length to be used (it was really short !). FAIL, it should accept any "reasonable" length screw without damage.
Did you not read the part about how it quickly gets separated from the manual? One can assume once it is separated from the manual, it is also separated from the little bag of screws.
Sometimes the mounting location requires a little imagination and the included screws would be too short since they've got to fit through spacers or something. Not everyone is going to professionally mount every bit of equipment in a standard 19" rack, you know.
The fixing screws go into the wall, and then the device is hooked over the screws. It's not clear how proud of the wall surface the screws have to be to foul the internal components of the device. I have a mini Netgear switch here with a similar fitting style. There is a insulated flap on the inside of the holes for the screw heads but this one is 12V only - the Cisco described takes mains.
Or how about idea for a protective mechanism?
Put a mesh of capsules containing oleum dulce vitrioli above the mains inlet so that any screw protruding too far will break a capsule, flood the area with anaesthetic and knock out any cowboy installers. I'm calling it "Ether Net over Power"
A lot of companies do this.. they grow so big they start to eat themselves from within and the right hand doesn't know what the left hand is doing. QA is the first to suffer with rushed dev schedules. (Microsoft is another example)
One bit of idiocy I personally experienced with Cisco switches in our organization was our units that had 2 redundant power supplies. If one supply outright failed, the unit would gracefully fail over. However, if the fan on one failed, the firmware was such that it would shut down the whole unit even with another perfectly good power supply available.
" If one supply outright failed, the unit would gracefully fail over."
That's ANOTHER problem in today's marketing-driven world ... only TWO power supplies? If you don't want the thing to break, n+1 is mandatory ... In other words, if two power supplies are needed to power the unit indefinitely, you must run at least three ... and they had better be hot-pluggable!
"However, if the fan on one failed, the firmware was such that it would shut down the whole unit even with another perfectly good power supply available."
'orrible, 'orrible quality control.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
