“Can you make my computer faster?” --> Can you stop making your computer slower? “How did I get a virus and how do I remove it?” --> An apple a day keeps the doctor away. “I forgot my password – can you help me get it?” --> No, but I can give you a new one. “Why won’t my printer print?” --> Why won't my ballpoint write? “Can I put more memory on my laptop?” --> Do you have a tiny screwdriver at home?
You should take a look on AutoHotKey. It can remap the keys according to your choice. If you want to assign the Win+# key combination to switch to desktop then you can use this script
#3::Send #d
Now when you press the Win+# key combination it will show up to your desktop.
# stands for Win ...
@avirk Just kidding. Altlhough, I wonder if you or the other person actually tried the hotkey. Because I wonder if it's even possible to override those shortcut keys.
Clippy's regexes are funny. Ask a question matching Clippy's input regex with "I'm" in it. It'll come out with "You'm" in the response. :D
I'd like to determine whether I'm trying to beat Clippy's regex system.
No, thanks. I already determined it.
To make Clippy's responses make sense, it would have to have a comprehensive database of irregular verb conjugations (including contractions) and conjugate all irregular verbs from the first person form to second person form
@HackToHell You have to ask something matching the regex /(?:^|[.!?:]\s+)(?:(?:how\s+(?:can|do)\s+i)\s+([^?!.:]+)\?|(?:i(?:\s+want|(?:\s+am|'m)\s+(?:wanting|trying)|'d\s+like|\s+would\s+like)\s+to\s+([^?!.:]+)(?:$|\.|!)))/i
Steps to reproduce:
Go to http://superuser.com/review/first-answers or any of the other Review tabs
Review posts
Reload the page
The posts are still there
Something broke and it won't remember the posts I've reviewed. This applies to all "old" review routes.
My computer at home is being a pain in the butt. I ran Automated Recovery and it found some corrupted .sys files and fixed them, but then it still couldn't boot (and didn't even get as far as it did before). Ran the wizard again and said that ntfs.sys was corrupted but it couldn't fix it. I also ran two full scans of my disk sectors and filesystem and it didn't detect anything wrong.
Note: This is for reference to point users who ask questions that will incite recommending software. I DO NOT advocate users asking for software recommendations, however there are times that a question requires the recommendation of a piece of software. If you're looking to answer a question tha...
@sidran32 hmm... and going from the installation CD, are you able to launch the recovery console and run sfc /scannow ?
you can't trust any of the files on the HDD itself as being valid so doing such diagnostics where the program diagnosing is on the HDD also is not acceptable...
it's like making a copy of a piece of paper with text then scribbling random magic marker on both the copy and the original, and trying to recover the original text
you can make N copies but all of them are equally susceptible to corruption
@allquixotic I did the "fix startup boot options". That's what I meant when I said the wizard. When two tries didn't fix it (the second one failing), I opened up a recovery console from the CD and did scandisk /x/b k: (IIRC, that was the switches I used)
@allquixotic It would make sense that it didn't work, though. Before things went completely to hell (and only was at the tipping point where it booted about 25% of the time), I did a full virus scan and it found and removed a bunch related to some Java exploit(s).
It wouldn't surprise me if I got hit by a virus that messed up my Windows install
Ah. I'll try that one next, when I get home later tongiht.
xD, my dad got hit by that too. Java 1.7.0 Update 5 was vulnerable to a zero day exploit allowing the execution of fully trusted Java code from an applet, which made it extremely easy to compromise Windows OSes (because from fully trusted Java, you can execute native code)
I removed it using Kaspersky's bootable Linux virus scanner
MSE removed it for me, that one time I got Windows 7 to boot. I have since uninstalled Java on all my machines, so it must have hit it before I did so.
fyi Java 1.7.0 Update 7 fixes the vulnerability, and if you're already infected then the virus doesn't depend on Java because it switches to native code once it deploys its payload
it uses Java (Update 5) once to get into the system; after that, having Java or not is a moot point
in my experience, the virus payload installed on my dad's PC was basically false security software, or "scareware", which intentionally disabled large swaths of the system, but without making it completely unbootable, and telling the user that you need to buy an expensive subscription to resolve the problems
but since the exploit allows arbitrary code execution, yours could be something that just flat out tries to make the system unbootable rather than keeping it nominally afloat but disabling most user actions
I know. I think it did its damage before I uninstalled Java. I ran the full system scan after Java was uninstalled and it removed the stuff.
I have no idea how it got on there, though. Usually I'm pretty careful about what I let run on my computer.
The first time it was begin flaky, I checked the boot logs (I enabled them) and the three system files it couldn't load were to do with the networking subsystem, antimalware, and windows file sharing. The first recovery fixed those (apparently) and after that it couldn't boot at all. The second recovery says that ntfs.sys was corrupted and it couldn't fix it.
It looks like it could have been caused by the virus, to me.
I just am hoping I can restore things without having to do a full system wipe. I already tried recovering from a backup, but the backup is also apparently affected by the virus.
I don't know how far back in time it goes and I don't want to lose too much by going too far back anyway. Maybe I can do an in-place install of Windows?
you're gonna have to put your waptop's HDD into an enclosure and boot up ubuntu live cd on the desktop and copy files over from the laptop hdd to the raid array... because you can't mount the raid array in the laptop
orrrrrrrrrrrrrrrrrrrrrrrrrrr just put the files on a flash drive and, again, boot up ubuntu on the desktop
and leave the laptop HDD where it is, nice and safe and cozy and protected in a nest of twigs and cottonballs and feathers and chicken wire and ladybug shells and nutella :3
/me delirious I need a day off... o wait I'm taking vacation tomorrow to board a train and I have to leave home at SIX FIFTEEN IN THE GOD DAMN MORNING.... jeebus... even my vacations deprive me of precious sleep
You know, to run the recovery console off the Windows install disc, I have to point it to my RAID drivers on a USB flash drive. So I can just stick the files on the same drive and use the recovery console to do the dirty work.
i think when i get home -- if i don't run off the side of the road and die on my drive home -- i'm going straight to bed, and not even packing anything into my suitcases or anything, just... get home, throw shoes in a random direction and immediately lose consciousness on my bed
and will wake up at like 4:30 and pack and get ready
well it can allow patent authors to more easily search for what prior art is already out there, and use that to tweak the wording of their patent until it just barely gets under the existing prior art
it's basically a cat and mouse game to come up with a random sequence of words that paints the picture of an idea or a product that nobody else has ever implemented or thought of before, and you get an absolutely insane amount of rewards for doing so...
The prior art is actually something the examiners have to look for. The patent authors rarely really research it in depth (I think, I'm talking out of my a** here)
judging from the painfully obvious things they've allowed to pass as inventions, there's no possible justification for those becoming patents other than their absolute stupidity
I think all patents should be abolished, and chiefly want to support/drive efforts to directly work towards that; but abolishing software patents would remove a lot of pain that I directly face on a daily/weekly basis as a guy deeply invested in IT, software development, open source, etc
trying to find prior art for new patents (because you only have 6 months' window after the patent is published where you can find prior art and submit it as a member of the general public) is a quest far, far, far removed from trying to abolish patents or even software patents
it's like trying to solve pollution by making plastic sporks out of paper
it would help reduce the petroleum material input to a very tiny, practically insignificant subset of the world oil consumption, but it would also be destroying trees, which combat pollution, and you still need energy (read: oil or coal) to produce them no matter what they're made of
why do you think software deserves special treatment? it's easy to get into the trap where you think non-software patents are OK just because you aren't directly affected by them
if you can understand why software patents need to be abolished, you should be capable of understanding why all patents need to be abolished
it's a stupid, childish game of "I was here first" that is only counterproductive and only promotes, in the end, despotism and tyranny
monopoly is just a thinly-veiled capitalistic precursor to tyranny
and the primary stated purpose of patents, even by their proponents, is monopoly
so you have a government-regulated tyranny whereby it's OK to have tyranny in the physical world over things like medicine and hardware and appliances and engineering and the physical sciences, but it's not OK to have tyranny in the virtual world over things like algorithms and software implementations
I think such a world would be very strange indeed and quite skewed if software patents were abolished but the rest were not
better off? perhaps... but the deeper problem would still remain
the problem with the patent system is that the definitions of the words we use are so contentious and so easily manipulable to allow patents to pass through any given filter (or in other instances, not pass through the filter) that a properly-greased gear could permit generous interpretations that would allow any rationally-derived filter to become useless over time
a "filter" such as "you can't patent mathematics" has already been proven to be worthless due to the huge number of software patents that are basically patents on mathematics
so if you make the filter more stringent, all you're doing is making it bigger -- but you aren't making it impossible to simply walk around it. it's like building a jersey wall to prevent people from getting from one side of the street to the other. it's a finite wall; eventually it ends, and at that point the person can just walk around it
so if you say "you can't patent mathematics AND you can't patent software", someone is going to find a way to create a patent that software infringes, which is not, specifically, a "software patent"
it's all about semantics and interpretation of definitions of words
by having a system based on that concept, any number of filters is basically just providing a long jersey wall for lawyers to find a way to walk around or climb over
so destroy your roommates until they accede to not using as much P2P traffic (reduce # of simultaneous connections etc), or install fq_codel (QoS won't help, by the way)
According to @balpha, he SE chat sites may have individual Easter eggs, depending on the site. If you find them, please post.
Edit:
Because it's so easy to make this stuff up, a screenshot as proof would be nice. (Although Photoshop ain't that hard either...)
when most people talk about QoS what they really mean is they don't want their internet downstream to grind to a halt when people have a lot of simultaneous / busy downloads going on within the same network endpoint
you want fair queue controlled delay active queue management for that, NOT QoS
you are suffering from buffer bloat due to the flawed design of the TCP/IP stack from the past decade
also I'm not quite clear on what the purpose of the DMZ is... if you want a single NAT subnet then you can plug in all the "child" routers/APs as simple switches and don't do any routing functionality on them
Hmm, blame it on a limited understanding of the overly simplified OSI model. I thought switches were only MAC-aware
Anyhow, the real problem is I have no control over either device right now ... the wireless router belongs to roommate abusing P2P, and can't reliably flash firmware on the AT&T device. Do you know if I can I install fq_codel on the RT-N16?
it's a kernel module in the mainline Linux 3.5 kernel (and later), so if you can boot your device on that kernel or the 3.3 kernel, then the answer is "yes"
iirc Dave Taht from bufferbloat.net is maintaining a port of dd-wrt or something like that which backports the module to 3.3
fq_codel has the interesting property that it "helps" combat the bufferbloat problem for all data flows passing through the device that it's installed upon
so for example, if you install fq_codel on your desktop, it will help combat buffer bloat if you have a download going on on your computer and try to surf the net... your surfing responsiveness will increase vs. not using it
if you install it on a router one level upstream, then any devices connected to that router will have the fair queue regulate the buffer bloat and those particular connections won't contribute to bloat and responsiveness will be improved
obviously the ideal solution is to have fq_codel installed all the way at the very end of the LAN chain, i.e. the network gateway, your modem
that way there is no traffic on your network that can get out to the public internet without going through codel
as long as you don't have some other connections coming "around" the RT-N16 and hitting the modem, everything will be "coddled"
so that's still pretty good
the modem itself will have bloated buffers but the important part is that the P2P traffic will actually be throttled, in a way, in order to allow other shorter-lived data flows to sneak in
visualize fq_codel like this: without it, you have a 4 lane highway packed solid with cars doing 85 mph and an on-ramp with cars sitting there idle waiting to get on the highway. there's no space and they're going 0mph and the traffic is going 85mph so they can't get on.
fq_codel adds a stoplight to force the oncoming traffic to give the on-ramp traffic a chance on in a while
as for your network architecture: if you've actually tested those DMZs and they work, great... but I'm not entirely sure that's the way to go... technically if you just don't use the WAN port of the downstream router(s) at all and just ethernet them into the LAN of the modem, your computers will receive DHCP from the modem
everything should automatically be packet switched and you can disable the NAT/DHCP functionality on the downstream routers so that they act as a dumb packet switch
so that when your computers do a DHCP broadcast they'll be hitting the modem and be on the same NAT
yeah it probably would be because your problem is about specific makes/models
a general question would be like, "another person on my network insists on saturating the connection with P2P traffic which kills my internet surfing responsiveness. is there a networking hardware/software solution that can regulate the traffic and improve my responsiveness?"
but I'm 99% sure that'd get closed as an exact duplicate since I've answered questions like that before
the answer is "Yes, use fq_codel"... by the way, you know Vint Cerf? one of the original designers of the IPv4 stack? he "endorses" the quest to fight bufferbloat and likes the CoDel algorithm
he's a super high level exec at Google and he's all over this stuff. \o/
I should add that that article is from February 2012, and they hadn't yet "discovered" Controlled Delay AQM by that time. CoDel is literally... like... born circa June/July 2012
extremely new discovery, but already in the mainline kernel and being evangelized by network engineers in the know ;)
they were still looking at SFB back in February
all I ask of you is that you try fq_codel (at least on your desktop Linux box or in a VM, if not your router), observe the effects with and without (e.g. with this test), and if you find that it works for you, pass on the word... evangelize it, make sure others know about the problem (bufferbloat) and the current best-known solution (codel).
hopefully within a year or two it'll be standard and we can resume saturating our network connections without seeing ridiculous and unreasonable latency added to the line
Given this question: Why did dialup modems make noise? I thought it would be fun to replicate this with a more modern internet connection, in that when the computer is connecting to a network (being wireless or whatever) it would play the dial-up noise.
A quick google didn't bring up anything ob...
I work at an ISP and we regularly get these questions, to wich i did not have a sufficient answer, until i went in to this chat room. Thank you so much!
@MainframeX Actually I would caution you to research controlled delay active queue management as well as buffer bloat some more before you take my advice
I'm not a network engineer or researcher, I'm only conveying my (interpreted) summaries of things I've read
since you work for an ISP, I would hope that you would do some testing and research on your own before making any decisions or giving advice based on this info
I don't know for a solid fact that all these problems are caused by buffer bloat, nor do I know that CoDel will sufficiently address them. but from my testing in my own personal environment I can vouch for CoDel certainly addressing the problem of latency spikes during uplink/downlink saturation.
I would actually be really thrilled if ISPs were to apply pressure to network equipment vendors (Linksys, Motorola, etc.) to ship CoDel or a close relative thereof as enabled by default on consumer routing equipment
that would be tremendously beneficial on the whole, I believe.
Sometimes customers (ab)use these lines to set up VPN's and citrix sessions and the like and go completely nuts when somebody starts downloading something.
also, the most important part about buffer bloat and anything promising to address it is that it only occurs when network utilization is at 100% within the endpoint device/gateway/router
simple logic: is network utilization 100%? No? Then buffer bloat doesn't exist in your situation, and any solutions fixing buffer bloat ain't fixing your problem.
and I don't mean link layer utilization... a DSL modem with a gigabit LAN can be 100% utilized to the internet even if it's only pushing 7 Mbps
but I assume you know what I mean by utilization since you work at an ISP :)
Spread the word to your colleagues/managers... test... and hopefully within a year or two this thing will have spread like a viral video and everyone will be happier. I honestly have no clue why the people who worked on codel haven't spread it to the likes of Slashdot, Reddit and C|Net already.
Step 1: download a file over HTTP as fast as you can from a server that can push more upstream than your connection's downstream. Step 2: try to navigate somewhere. Step 3: throw your computer out the window (pre-CoDel) / Step 3 w/ Codel: enjoy the internet low-latency and all ;)
I mean I wish you had already known about it etc ... it's just one of those weird things where the researchers working on it haven't gone to the news and said, "Hey, you should really look at this!" -- it's really remained under the radar, despite being in the mainline Linux kernel. O_o
All these years people google around for how do i stop my internet from grinding to a halt when i download something, and all these years people come up empty or get some vague reference to QoS or something
that used to drive me up a wall when I was sharing my home internet connection with my dad.
he'd stream a video and my online FPS game would just be like... "Connection Interrupted"
at one point my dad and I had Verizon ship us five different models of ADSL modem attempting to fix that very problem
but nobody had any clue what the issue even was in 2008....
I ended up installing a proprietary software program on my dad's computer called NetLimiter which capped his downstream at some fraction of our DSL's downlink rate
that did remove the saturation problem, and in turn, got rid of buffer bloat -- but of course, I never made the mental connection of what the actual issue was
we went through that process about five times before just saying, you know what, we're going to do this our way, and that's when I put NetLimiter on my dad's PC
I put it on mine, too, to keep my downloads from ruining his experience
so we ended up getting roughly 60% of our DSL's downstream for any individual download just to leave some available bandwidth for the other guy
CoDel lets you hammer your connection, outright saturate it, and still get a good experience... so that's why I like it... you no longer have to prevent saturation from occurring, which basically surrenders to the buffer bloat problem by saying "I'm just not going to use all the speed I paid for".
it must really suck if you're on something like fiber 100 Mbps and you have to limit your downstream on that expensive and crazy fast line because all uplink speeds suffer from the same problem when saturated :P
I originally thought it had something to do with the fact that ADSL is asymmetrical, because TCP protocol requires that downloading packets have to be ACKed on the upstream, and it's possible for you to flood your upstream with data and not leave any room for the ACKs, which slows down the downstream, etc
but then I saw that the same symptoms were occurring when nobody was "uploading" anything (other than, you know, HTTP requests and ACKs, very small, low-bandwidth data)
I'm going to send a link to the ACM article about buffer bloat (the one with Vint Cerf and Jim Gettys) to my friend who worked at Cisco for 30 years, and now works for some other networking company... see if he can find any merits in it... or knows about it already...
I only found out about it myself very recently
it's time for me to go home. but hey, do post back on this channel if you and your work buddies obtain some results from testing CoDel :)
You should take a look on AutoHotKey. It can remap the keys according to your choice. If you want to assign the Win+# key combination to switch to desktop then you can use this script
#3::Send #d
Now when you press the Win+# key combination it will show up to your desktop.
# stands for Win ...
Steps to reproduce:
Go to http://superuser.com/review/first-answers or any of the other Review tabs
Review posts
Reload the page
The posts are still there
Something broke and it won't remember the posts I've reviewed. This applies to all "old" review routes.
According to @balpha, he SE chat sites may have individual Easter eggs, depending on the site. If you find them, please post.
Edit:
Because it's so easy to make this stuff up, a screenshot as proof would be nice. (Although Photoshop ain't that hard either...)
