09:18 – We’re having our deck replaced. Yesterday, a truck showed up with the framing lumber. This morning, the work crew showed up to tear down the old deck. We had just under five inches (~12.5 cm) of rain last night, but our back yard is on a grade so it’s drained well enough for them to start work.
We’re nearly ready to start shipping forensic science kits. I have a couple more chemicals to bottle and then we’ll start building subassemblies. Until we get all of the subassemblies ready to go, we won’t know what size box we’ll need. I’m sure everything will fit in a USPS Priority Mail large flat-rate box, but I’m hoping it’ll fit a PM regional rate box instead. If so, that’ll cut our postage costs by an average of maybe $3.
I’d intended to drop our wired phone service and go 100% cell. But thinking about it this morning, I decided to sign up for a third-party VoIP service. I looked at Vonage, but I decided to reactivate our service with PhonePower, which we had before we went to TWC VoIP service. PhonePower service was great for a few months, but then it started hanging unpredictably. I think the problem was the terminal adapter. The one we had back then had to be on the local side of the router. The TA we’re getting this time sits between the cable modem and the router, so there shouldn’t be any problems with hangs.
16:39 – When we’re making up chemicals for kits, I invariably save the most obnoxious chemicals for last. This time was no exception. I just finished making up a batch of Kastle-Meyer reagent, which is used as a presumptive blood test. Actually, K-M reagent isn’t as obnoxious as some. It has no odor. The obnoxious part is having to reflux a 40% potassium hydroxide solution, which literally eats glass.
I originally intended to make up a 2-liter batch, but I decided one liter was sufficient for this pass. That’s enough for 30 forensic science kits, which is the batch size we have in progress. So I halved the recipe, transferring 500 mL of DI water to a 2 L Erlenmeyer flask, dissolving 200 g of potassium hydroxide in the water, adding 20 g of phenolphthalein powder and 200 g of zinc powder, and then refluxing it for an hour or so, until the bright pink solution turned colorless. That’s cooling down now. When it’s at room temperature, I’ll make it up to 1 L with 70% ethanol and bottle it. Some sources say that K-M reagent is good for months stored in a sealed bottle, or years if it’s refrigerated. My experience has been better than that. I have some K-M reagent I made up in 2007. It’s been stored at room temperature, and it still works the same way it always did.
APC 3KVA UPS at work popped and belched the magic smoke. Called APC and they said the unit is not worth repairing as it is over 8 years old. New UPS on order. Servers meanwhile are on a much smaller UPS that has barely enough capacity to handle the load. I used it because I needed the outlets. New UPS is on order.
Took the unit apart to see what blew suspecting capacitors. Nope, it was the rectifiers. Six of the 18 rectifiers are gone, holes in the side. At the university they have been working on replacing the power busses in the building. I suspect, but cannot prove, that the university crossed something and sent 480V in my 120V circuit. I suspect this because the lights flickered twice during the event. It is a dedicated circuit and the wiring in this building is whacked. We have circuit breakers on the 4th floor, 5th floor and 6th floor for our offices on the 5th floor. I have no idea where my server circuit (20 amp dedicated) terminates and UT probably does not know either.
In 1995 I was doing some consulting for GE Fanuc, an engineering
consulting company. Work crews elsewhere in the area spiked some power
lines, crossing the 400V 3-phase and the 120V, or something similarly
destructive. The power in the building went out and stayed out. Bad. The
array of AIX servers lost power. Bad. The AIX servers had no UPS. Worse.
The AIX servers suffered hardware damage because they’d just been
plugged into the wall with no surge suppressors or anything. Very bad.
The AIX servers’ disks had not been backed up, ever, because it was no
one’s job to back them up. Catastrophic.
If I were looking to hire engineering consultants, GE Fanuc would not be
on the short list.
We are backed up quite well as that is part of my job. The UPS took the brunt of whatever happened. It could have been one rectifier that blew and the rest all went in a sympathetic gesture.
Memo to self. I need to check the voltage on the outlet.
Upgraded to Firefox v15 this morning. It has cut WAY down on RAM usage. I currently have about 60 tabs in 4 windows open and my total RAM usage (for the whole computer) is still under 1gb. Unheard of since about v3 or 4.
I usually only have one instance of FF (14.0.1) going with 30-40 tabs. It’s been hanging a bit lately and I got a BSOD yesterday while FF was thinking about something.
That’s a LOT of porn. Though, I suppose variety is the spice of life. 🙂
Oh, normally I have at least 15 Firefox windows open with a minimum of 10 tabs in each. My son has run Linux for about 8 years. He uses terminal commands and does not run a desktop. He had a 60gb hard drive, and a few months ago, I gave him my old 500gb drive. He somehow suspends in Linux, so he has never actually shut the machine down—it saves the configuration to hard drive and opens everything up to where it was when you last quit.
When we were in the process of swapping drives, it took him 2 hours to dispose of over 200 windows he had open. I could never keep track of that many windows, but there are Linux windows managers that are built for people like that. Even so, he said it was a nightmare getting anything done for about 2 weeks, until he found everything that had to be closed.
As regards:
Unfortunately “shouldn’t” is indeed the operative word. Bufferbloat in the cable modem could still cause hangs. That’s where the makers of the cable modem decided to include an extra-large buffer, memory these days being cheap. When you do a large upload, and the cable modem’s buffers are full of two seconds’ worth of data, your voice packets get delayed by two seconds, which ruins the call quality, if it doesn’t kill the call.
There are mitigation strategies that work pretty well for bufferbloat, but the first step is to stop by Netalyzr:
http://netalyzr.icsi.berkeley.edu/
and see what it reports regarding buffer sizes, to see whether you have the problem to begin with. Or for more background, see:
http://www.bufferbloat.net/
Or do like I do and run two modems, one for internet, one for voice. Comcast only charges me for the one modem.
Wouldn’t the TA prevent that from happening by taking precedence. I haven’t noticed any problem using my Ooma box. It’s located right after the modem. I mainly download, but the box has never dropped a call with 5 people surfing, netflixing, etc. at once.
This is why our educational system sucks: because we have people running it like this high school principal who withheld the valedictorian’s diploma because she said “What the hell?” in her graduation speech.
http://www.theatlantic.com/national/archive/2012/08/terrifying-teen-speech-in-the-news-again/261534/
When I first read about that valedictorian, I wondered why the school board didn’t crucify the principal. What a dumb and arrogant thing to do. That’s what you get with “zero tolerance” policies. Nobody can think for themselves.
With bufferbloat, uploading is usually the problem: the upload channel is lower bandwidth to begin with, so if they make the buffer the same size in both directions it’s too big for upload. That’s when RBT’s previous VoIP most notably got sketchy: when he was uploading videos.
As for the TA “taking precedence”, it can’t do much to control the cable modem. All it can do with packets coming up from the router is to either drop them or pass them on, and without knowing the upstream speed it wouldn’t know how to ration packets so as to give its own packets enough room. It could set the “quality of service” bits in the IP packet headers, but the cable modem will probably happily ignore those.
(Determining the upstream speed is one step in mitigating bufferbloat, but it’s hard to do automatically.)
No, my problem with the Windstream TA had nothing to do with upload traffic. The TA would simply stop working randomly and require a reset. That was particularly problematic given Barbara’s parents’ situation, because we never knew if our phone was working or not unless we picked it up to see if there was dialtone.
I actually thought about seeing if I could use our current TWC cable modem’s TA. It’s a special model that they install only if a customer has VoIP service from them. I suspect it won’t work with the phonepower service, but I’m going to look into it in my copious free time.
One thing you can do with Phone Power is set up an alternate phone and if the TA is unreachable it will send the call to that. In my case my cell.
Well, you did report problems when uploading videos:
http://www.ttgnet.com/daynotes/2010/2010-34.html#Tue
Uploading a video shouldn’t really kill other uploads; that was likely (although the term hadn’t been coined yet) bufferbloat at work. Internet flows are supposed to play nicely with each other, only causing occasional packet loss when they exceed a connection’s capacity. And with VoIP, occasional packet loss results in glitches in the audio, not the thing ceasing to work entirely. But the control algorithms that enable flows to coexist don’t work well when there are multiple seconds of buffering in the loop.
As for using the current cable modem’s TA, no self-respecting monopoly would let you do that.
It looks to me like Phone Power includes two lines. If I were you, I’d be inclined to forward all the calls from one of those lines to at least one cell phone. Give the number of your forwarded line to Barbara’s parents and other people you really want to hear from. Give the unforwarded line to everyone else.
Two lines on one number. The TA has two phone ports but if you want a true second line with separate number you would have to pay for that. Right now I have mine attached to a two line desk phone and what happens is it rings the lines alternately when someone calls in. You can put one line on hold and answer a second call or make a call out. You can use the phones conference feature and tie two calls together. It’s still one number on both ports. On incoming mine only tries for about 10 secs before I have it forward to my cell. If I’m not sitting there there’s no need for a long wait to forward.
The have all kinds of calling rules you can set up on line.
So far I’m pretty happy but after AT&T who wouldn’t be.
At the new house we are apparently gonna be stuck with Comcast, so we are only getting the fast internet from them, no tee-vee and no phone. We will use Skype for long-distance which we do now anyway, and the cells for on the road, backup, and emergencies, with chargers by the bed.