and now, not politics but technology

[rmd]

so, when last we left our intrepid network engineer, she was grappling with pesky 20th century technology of 10M multimode fiber.

after two next-day-delivery orders from blackbox (i mistakenly ordered 1300nm MMF converters the first time -- OOPS), and way too many phone calls and emails where i tried to be the ringmaster of a three ring circus of vendors all saying "NOPE, OUR STUFF IS FINE. IT'S SOMETHING ON THE OTHER SIDE. OR MAYBE THE OTHER OTHER SIDE," i am finally left at this state:

i have an allegedly properly-working canoga perkins mux turning my wan circuits into 10M/full-duplex MMF on 850nm. i have a cross-connect fiber infrastructure leading from the cage where the mux is thru a couple of patch bays to my cage where i can *finally* see that 850nm of red coherent photony goodness coming out of my patch panel. i have an lc/sc mm cable connecting from the patch panel down to my blackbox lmc002a-r5 media converter. i have an ethernet cable leading from the media converter to my router.

and now, link status on my end and link status on the canoga perkins are related by an XOR.

which is to say, either i have link on my side, or they have link on their side, but not both. they've tried swapping cables around and using one strand from one cable and another strand from another, but, well, no change.

the colo tech was as baffled as i was. i left him last night to try another pair of fiber under the floor in case there's something wackily broken in that.

i'm just stumped by this.

WHAT THE FUCKING FUCK, PEOPLE.

and all this for a goddamned 10M link. i have a theory: the slower the link, the more of a fucking pain in the ass it is. so far, this theory is supported.

Comments

[blackanvil.livejournal.com]

Yeah, we used to dread 56k frame relay ciruits, because they were the hardest connections to get up and running -- mostly because nobody cared if they worked or not except the customer. Oh, wait, there was one type of circuit worse -- 0 CIR international satellite frame circuits. If we could get even a 10% ping rate through on one of those, we considered it working well enough.

[be-well-lowell.livejournal.com]

i have a theory: the slower the link, the more of a fucking pain in the ass it is. so far, this theory is supported.

With extra pain-in-the-ass points for the length of time it's been since a technology got a new customer.

Says the guy who needs to get ATM APS running on a modern provider-edge router before I can move on to fun things.

[andybeals.livejournal.com]

Ok, so you have eye-frying goodness coming out of the fibre. But do you have *link* according to the GBICs? You might have to do a left-right swap on the fibre, as there's no standard as to what side goes to which end. (Or if there is, even the wiring vendors fuck up.)

[rmd.livejournal.com]

i'm sending light. the far end is sending light. it's visible light, so i know we're not screwing up the tx/rx strands. but only one end will claim to have link at a time.

so weird.

[andybeals.livejournal.com]

Ok, so either the fiber's horked or someone's got a bad GBIC. Fenisars? Even better if they're the same manufacturer/model (exact model code printed on the little bugger).

[rmd.livejournal.com]

yep. that's what i'm leaving the colo folks to test out. i'm hoping it's something simple like something wacky with the underfloor fiber feed.

[xthread.livejournal.com]

the slower the link, the more of a fucking pain in the ass it is

This is observed to be empirically true.

Three Ring Circus

[catness.livejournal.com]

Okay, so... part of my job is to do what you're doing on a fairly regular basis. My territory is Chicago, even though I've never been there, and we just had a turnup last week. In the process of getting everything done, there was a lull in my Chicago staff's workload, when my newest guy said, "So, we've hooked everything up and we seem to have link, but nothing pings. What's next?" My second-most senior staff replied:

*** Topic for #teamchicago: next step: everybody says its everybody else's fault

So, while I sympathize, I'm also laughing a lot.

Re: Three Ring Circus

[rmd.livejournal.com]

bwahahaha.

also, that user icon is pretty awesome.

Re: Three Ring Circus

[catness.livejournal.com]

You are quite possibly the only person who has seen it and found it even slightly relevant to the universe. :)

[madbodger.livejournal.com]

I now understand your fondness for OC48 circuits.