This forum is in permanent archive mode. Our new active community can be found here.

GeekNights 20110328 - Logs

edited March 2011 in GeekNights

Tonight on GeekNights, we consider that most basic of basics: the humble log file. In the news, Geohot is fighting the good fight, and is the financial world siphoning off all of the world's talent?

Download MP3
«1

Comments

  • My best score: 14.2 m
  • My best score: 14.2 m
    How about at the prequel QWOP?
  • edited March 2011
    3.0 m
    Post edited by Ikatono on
  • edited March 2011
    Rym's Thing - Toy Snake Math
    I didn't think it was possible to fall in love with someone so quickly. This woman has captured my heart in so many ways.
    Post edited by Victor Frost on
  • GIRP, my best score? 72 meters. But I can't reach the present! It makes me so mad.
  • Scott: You're a smart guy, but you have no idea what you're talking about when it comes to Video Editing.
  • The talk of video editing is making me twinge, however like most other people everyone has to learn somewhere.
  • edited March 2011
    The 720p version of one of the 15 minute clips downloaded from youtube is 135MBs, not 700. Also, a random video-editing anecdote, bloopers are best set to Drinking by TMBG.
    Post edited by Pegu on
  • The 720p version of one of the 15 minute clips downloaded from youtube is 135MBs, not 700
    Hahaha... So, funny story. I download them back from Youtube to see the difference. I then proceeded to look in the directory I'd stowed the original renders when I went to do the comparison. Magically, they were the same size. ;^)
  • OMG. They mentioned one of clearest memories of a computer game I played as a teen. Phantasmagoria! I still have one of quotes burned into my mind. The scene is you walk into a room where your husband is and there is 100's of pictures of you and your head is ripped off of the photo. Your husband says, "...."
    I'll post the quote if no one knows it.
  • I can't stop playing GIRP. I guess you could say it has me in its...grip.
  • edited March 2011
    Ready for some snake math?

    The smallest shape that will make the snake hit itself without going straight is RRR/LLL, or three of the same turn in a row.
    The next smallest is the outline of a plus shape (RLLRLLRLLRL) or its inverse, or a reflection of either. However, this requires 11 joints, so any combination is good as long as it doesn't have 3 of the same turn in a row.

    I worked out the simple values and got the following results:
    1: 2
    2: 4
    3: 6
    4: 1
    5: 16
    6: 26

    This appears to follow a Fibonacci-style sequence, so I extrapolated the following answers:
    7: 42
    8: 68
    9: 110
    10: 178

    I don't have a proof that that's the right pattern yet, but I recently did a similar problem (find the number of intergers <2<sup>n whose binary expansions don't have two adjacent 1s) and it had a similar answer, so I'm feeling confident. I will try to work out a proof, though.

    I'm not even going to try solving it if they can be straight right now, that's much harder.
    Post edited by Ikatono on
  • Here's my proof:

    Let's assume that the last joint on n-1 is R. The next joint (the last one of n) can always be L, so you have f(n-1) possibilities. The next joint can be an R instead if the second to last joint was an L; otherwise it would form RRR. There are f(n-2) possibilities in which the second to last joint is L, so the total is f(n-1) + f(n-2). Thus:

    f(n) = f(n-1) + f(n-2)
    f(1) = 2
    f(2) = 4

    This will work for anything up to 10 joints, after that you run into more problems.
  • That seems correct.
  • Right now I'm working on expanding it past 10 joints. The shapes that are no good are a square, a plus sign, two plus signs connected at a side, three plus signs in a line, four plus signs in a line, etc. so i think there's a formula to solve it for any size of n.
  • edited March 2011
    Just note that with 11 joints, there's 6 ways to get a plus sign:
    RRLRRLRRLRR
    RLRRLRRLRRL
    LRRLRRLRRLR
    and the same ones with the Rs and Ls swapped.

    Alternatively:
    A complete plus sign is cyclic, consisting of 12 joints of the form 4(RRL) or 4(LLR). There are 3 distinct ways to take a single joint out of such a cycle in order to make an 11-joint snake, and we have 3*2 = 6.

    I'd say that f(11) = 178 + 110 - 6 = 282
    Post edited by lackofcheese on
  • edited March 2011
    Right now I'm working on expanding it past 10 joints. The shapes that are no good are a square, a plus sign, two plus signs connected at a side, three plus signs in a line, four plus signs in a line, etc. so i think there's a formula to solve it for any size of n.
    It's much more complex than just plus signs in a line; there's lots of ways to make a snake crash.
    Consider this 15-joint configuration:
    RLR LLRLLR LR LLR L

    Or, as a 16-joint cycle:
    2(2(LLR)LR)
    or
    2(2(RRL)RL)

    There are 8 distinct places to break one of these cycles, and so there are at least 16 different 15-joint configurations that crash while making neither a plus sign nor a square.
    Post edited by lackofcheese on
  • Also, I was disappointed to find this episode was not about logarithms, nor about branchless tree trunk segments.
  • The talk of video editing is making me twinge, however like most other people everyone has to learn somewhere.
    It funny, it's like watching a little kid grow up. Awww, who's working on video now, who's a widdle filmmaker.
  • The talk of video editing is making me twinge, however like most other people everyone has to learn somewhere.
    It funny, it's like watching a little kid grow up. Awww, who's working on video now, who's a widdle filmmaker.
    You should promise to take them out for ice cream after the last clip is uploaded.
  • It's funny that our positions are a little reversed now.
    Emily went to film school, she now works on videogames and is learning programming, Rym went to tech school, he is learning film.
  • It's much more complex than just plus signs in a line; there's lots of ways to make a snake crash.
    Consider this 15-joint configuration:
    RLR LLRLLR LR LLR L

    Or, as a 16-joint cycle:
    2(2(LLR)LR)
    or
    2(2(RRL)RL)
    I missed those completely. I didn't spend too much time on >10 joints. I would probably work on this more, but I need to get to my PROMYS application, so I've got plenty of math for now.
  • edited April 2011
    So I've just gotten around to listening to the whole episode and I'm in so much fucking agreement with Rym on how most IT people don't look at the logs. But I disagree about the usefulness about the Windows Logs.

    Let me set the scene.

    I work as a junior systems administrator for the Information Systems department in the College of Engineering and Computer Science at Cal State Northridge. My official title is student assistant, but I do just about everything any of my seniors in the deparment do as well as the majority of the manual labor, but I digress.
    About three for four weeks ago, we started having problems with Casper (WindowsServer2008), our Symantec Ghost Console server and department software repository for our college. Some of the disk images for used for imaging the lab computers were coming up corrupted, software installs were having problems, and in general, Casper was acting sluggish. Martin (name changed), my direct supervisor, is in charge of the managment of Casper. Me, along with everyone else in the office, have been on his back about the state of Casper for a while now, but he hadn't done a thing about it. Finally, the most important image on the server became corrupted and shortly after, the server went down.

    NOW he looked at it. After a week of being holed up in the server room he got it back online, but neither he, nor any of my other superiors could figure out what was wrong. Finally, Casper stopped responding and went down again and wouldn't come back up. User support type servers (like print and exchange) aside, Casper is the most important server in our server room. This was a major problem. Martin was out for the week, so I told the next person up, the boss of our whole department, that I was fed up and going to take a look myself. It should be noted that one of the big reasons I was hired was for my unconventional thinking and my ability to come up with new solutions to strange problems.

    So I head up to the server room, pull out the KVM drawer, and power cycle Casper. Casper is a six or eight U HP server with a five or six disk raid array, so we are used to it taking a long time to boot. A half hour is the norm. I went to go get lunch and hour later, it was just finishing booting up.

    That fucker felt shaky as hell, but I managed to log in and pull up the management console. I look through the system logs at the points of the two failures. Apparently, the raid array had reported a critical error with one of the disks and halted the system. On top of that, all logical drives on the raid were fragmented above 70%. So, I went back downstairs and explained what was up and told him what disk they needed to replace. He asked me how I knew and when I told him, he acted like he must have missed it, but from what he was saying, it was clear that he didn't even know WHERE THE FUCKING LOGS WERE.

    *Sigh* Sometimes I get the feeling I'm over qualified for this job. Or at least this title and paygrade.


    EDIT: What is that tool for viewing the windows kernel logs?
    Post edited by Victor Frost on
  • I think the Windows kernel log tool is DebugView.

    Actually, you may want to look at all the Windows Sysinternals utilities. Lots of useful stuff there in general.
  • look through the system logs at the points of the two failures. Apparently, the raid array had reported a critical error with one of the disks and halted the system. On top of that, all logical drives on the raid were fragmented above 70%. So, I went back downstairs and explained what was up and told him what disk they needed to replace. He asked me how I knew and when I told him, he acted like he must have missed it, but from what he was saying, it was clear that he didn't even know WHERE THE FUCKING LOGS WERE.
    Ahh, but those logs are useless if you're already monitoring the array (which you should be). If an array, or any monitorable hardware component, has a problem, my poll-based monitoring catches it within 30 seconds.
  • I think the Windows kernel log tool is DebugView.
    Basically. I have a slightly different tool a vendor provided, but it does the same thing.
  • Logging is you filling out the questionnaire when you get to the doctor's office. Monitoring is the clipboard the doctor writes on after poking and prodding. You gotta have both.
  • Logs? Last I knew, they're big, heavy, and wood.
  • Ahh, but those logs are useless if you're already monitoring the array (which you should be). If an array, or any monitorable hardware component, has a problem, my poll-based monitoring catches it within 30 seconds.
    How do we do this on Windows? I want this fucking thing. I want it on all of our servers.
  • Logs? Last I knew, they're big, heavy, and wood.
    I believe they are also known to roll over your neighbor's dog.
Sign In or Register to comment.