Minutes of the SIPB Meeting of 07/31/2006 The meeting was called to order at 19:30:01 by tabbott. In attendance were Voting members: jhawk tabbott mshaw aseering kchen | yoz Associate members: asedeno Prospectives: mathmike austein nelhage mitchb Guests: | sipb3 (Aman Gupta) Treasurer's Report: jhawk: I just wanted to remind the Board that there is this $1700 I'd like us to spend this month, and I'd rather it not lie around in "La La/jhawk" land for a long period of time. Chairman's Report: tabbott: election sometime during the meeting mshaw: I move to censure the Chair for starting the meeting 5 seconds late. tabbott: Motion failed for lack of second. jhawk: I'd just like to note for the record that the meeting started ONE second late. [ Election begins at ~19:46] tabbott: Introduce yourself. nelhage: I'm Nelson, username nelhage, uid 57530, a rising sophomore, I've been hanging around SIPB a lot and doing a lot of work on the Linerva project with tabbott, packaged a lot of the Athena stuff for Debian that we're finally getting around to releasing and tidying up. I'm a VI-3 major. Hi. tabbott: Bonus points for wearing a SIPB shirt today. jhawk: [Long involved question involving dealing MIT spam, how to improve the problem that's been much worse in recent days, what can SIPB do about the problem, both technically and socially, and might it be helpful to collect staistics on the problem, etc.] nelhage: I will start with the observation that I and many other people seem to be noticing is that it's technically possible to trap this spam. So either the spamasssin implementers or maintainers are not managing to solve the problem... I do not believe that spammers are simply getting smart enough. nelhage: In terms of social action SIPB could take, I don't really know what the situation is, I don't know the details on MIT's implementation. IT should be possible for SIPB to encourage MIT to take whatever action. Rather than ??? nelhage: There are probably parts of the question I failed to answer... jhawk: Yeah, like what you personally could do? nelhage: I'd have to take some time to learn more about the parties involved to get a better idea of what action. I would probably be interested in aiding an effort but not in leading it. mshaw: Nelson, if you look around this room, you'll notice something about everyone here. We're all male. This is a problem generally in computer science and MIT circles. It's often not this bad. We're generally... What you would do, as a member of SIPB, to help this situation. Or do you not think it is a problem? nelhage: I think it is an unfortunate problem. I don't know what the statistics are, but we seem to be far more skewed than the population as a whole or even the CS population; the room is 100% male, and even with active people like jtu and k_lai our ratio is still not representative. I'm not sure offhand what we can do to fix it. SIPB does have occasionally something of a reputation as a repository for sketchy old cruft with no other place to hang out. Possibly of concern to some female frosh. There are a lot of cruft who are members of SIPB and this certainly shouldn't change. I don't know what we can do to change that impression. nelhage: Possibly more active attempts to recruit people in some sense. Most members of SIPB end up coming over here because they noticed our tours, publicity, etc., and wander over here of their own accord. SIPB members who know female members of the MIT community could suggest that, if you happen to be interested in computers, come by SIPB. tabbott: So what new projects do you think SIPB should be doing? nelhage: Well, we should be rewriting emacs. We could launch a SIPB zephyr server so that when MIT takes theirs offline, we can simply transfer to the SIPB realm. We should be implementing a new more awesome owl that doesn't leak memory and is somehow more awesome. It should run under emacs. mitchb: I thought you just said it shouldn't leak memory. mshaw: This is the NEW emacs. jhawk: it's pronounced GNU emacs. ???: ... SIPBmacs. mshaw: What is the least valuable current active SIPB project? tabbott: What do you mean by active? nelhage: Yeah. mshaw: Something touched in the past two yars. For example, not SIPB-RT. [~20 sec pause] mshaw: We could start off easier. What's the most useful SIPB project? nelhage: Scripts is probably "up there." It's really useful for people who don't have the resources for their own servers, and as an incoming frosh I was frustrated that MIT didn't provide it, and I was excited when I found out there was a service that SIPB did provide. nelhage: Shall I go on to least? The last 2 years is unfair, I've only been here for a year. mshaw: Last one. nelhage: stuff.mit.edu is a cool idea, though it fizzled, and nothing really good came out of it. It might be nice to have a central web location customizable per user, but it is my impression that nothing actually significant came out of it. jhawk: So, tell us more about yourself. What one thing, that relates to SIPB, are you really the best at? nelhage: I'd have a hard time pegging any one thing... I feel that I'm very good at... gnerally fairly willing to dive into a wide variety or projects/challenges and go out and acquire whatever knowledge is necesary. jhawk: OK, that's nice, but a cop-out answer. Surely there is something you are particularly good at. yoz: The answer could be no. nelhage: I will pass for the moment. yoz: With the ever-changing role of Athena, what role do you see SIPB taking in the future with regard to computing at MIT? nelhage: I feel like we're increasingly taking is offering cool things that Athena as a project is unwilling/unable to offer. Things like scripts, like linerva, a public Linux dialup for the community; trying to move past the idea of this one Athena Linux [and Sun?] platform, anyone who has a Debian laptop... nelhage: Taking on cool useful services for the community that the Athena instituion is not tackling. That is a reasonable role for SIPB to take. To keep providing all this additional value on top of Athena. As well as doing whatever we think we can to work with groups like ISTAB to ensure that Athena doesn't become lame. I realize this is a hard thing because people are trying to make Athena lame *cough*WinAthena*. asedeno: On the topic of diving into things, and really getting at the root of certain problems, what would you say your least favorite and most favorite bits are, of sendmail, as implemented on Linerva? Or across other parts of Athena? nehlage: It's hard to pick a least favorite. Some of my least favorite are: it's running suid root; running with a sendmail.cf that was hand-written sometime ago without using m4 or anything that attempts to make it remotely readable; like a lot of things on Athena it's wrapped around by a neat script that sets the MAILRELAY environment variable instead of the user invoking sendmail direclty. These have all been great annoyaces to me at some time. tabbott: The build process does not win? nelhage: The build process is not one of my favorites. It's pretty bad. But suid root is pretty bad. One of my more favorite things is that it will do Kerberos authentication with your tickets, which for example postfix is just flat-out unable to do. I haven't tried extensively with exim or qmail but I'm not convinced that they can either. sipb3: Last time I was here, which was a year ago, how many of the machines in this room can you name? nelhage: quiche-lorraine, deathtongue, general-dink, lying-naked-in-the-periwinkles [sic], opus, bart-savagewood. jhawk: It's lying-naked-in-the-periwinkle. mathmike: Do you not sit in the hotseats enough? nelhage: I do not sit in the hotseats enough. My laptop over there is "phanatique". Oh, one of those is zygorthian- space-invaders [...] , err, zygrothian-space-raiders. jhawk: You're not actually going to provide a number? tabbott: I think that is unecessary. yoz: Suppose you are sitting in the SIPB office alone, and are prepared to close to be somewhere in 10 minutes, and a user comes in w/ a question about printing a web page. What do you do? nelhage: I would attempt to ascertain if it was a simple issue I could address verbally first, figure out exactly what his issue was. After that, if it was some complicated issue (I couldn't just tell him -Pajax and make him happy), depending on the urgency of where I had to be, I might take a minute to try to help him. If I had to be gone, I'd regretfully inform him that there was somewhere I had to be. If he was a zephyr user I'd direct him to -c help or -c sipb. Otherwise I would probably, if he continued to have the problem, he could email sipb@mit or try searching around MIT's web pages, though I don't know if there's any information that might be useful to the specific problem. jhawk: Something else you might do is mention OLC and the OLC stock answers. jhawk: [Long anecdote about Winathena, flash not installed, what does this mean for WinAthena, and what do you think of it in general?] nelhage: I have not used WinAthena myself recently, or had to help users with problems. It is my impression that it sort of just doesn't work well, it doesn't provide a lot of the nice things that a UNIX Athena does. The Athena aspects aren't nearly as well-integrated into the system. Umm. I think it is sort of unfortunate that it keeps becoming a larger part of the computers on campus. Although I am personally not a huge fan of the UNIX Athena as opposed to another Linux distribution, I think they are much nicer to work with than Linux boxes, I think it is useful for people to learn how to operate them. I think they are a much ??? jhawk: I had a user who came in and observed that the WinAthena machines are slower for running MATLAB than UNIX Athena machines... nelhage: I don't know what the WinAthena hardware in clusters. I expect it is to be expected. jhawk: Aren't economic issues relevant? nelhage: *shrug* kchen: I reinstalled a system that used to have Athena with Ubuntu and the machine is faster, but it has fewer users. I expect that not running stuff out of lockers makes a difference... nelhage: That most stuff comes out of the network, that could generally be a problem... mshaw: On the same topic of WinAthena, a lot of those machines are dual head. Do you think that this is a calculated move on the part of some folks to switch people over to Windows machines? nelhage: Quite possibly. mshaw: Is this something SIPB should be protesting in some sense? nelhage: It's my understanding of the Linux Athena machines (and Linux distributions in general) that there are a lot of issues getting dual monitors to work. If we did fix that, then it might be reasonable to advocate. mshaw: It's not always just a technical problem. A 24-inch monitor is about the same as two. aseering: I sat down and figured out how to make Athena dual head and the box that I did that on was replaced with general-dink. Do you have any objections to making general-dink a dual-head Athena box? nelhage: Is general-dink currently an Athena machine? tabbott: It has an Athena VM. I think we were waiting for you to do to the thing you did to the ESP office machine so it's useable. nelhage: I would suggest, dont' try to do it on general-dink, because it is running Xen. I personally think it would be awesome if the office had a dual-headed Athena machine. kchen: I think IS&T may have done this on one Athena machine, specifically the one on the back room of the W20 athena cluster. I know they did it so you can display the same thing on both the screen and the wall-mounted screen. mathmike: It seems doubtful. nelhage: My experience with Linux is that mirroring is a lot easier to set up. aseering: I've found it to be a lot simpler. nelhage: Ask someone in the room a question that only you can answer. mshaw, what did you have for lunch? [debate on merits of question] mshaw: You had food for lunch today. You never specified the specificities. tabbott: The question is re-asked with "technical" inserted appropriately. mshaw: Technically, what did you have for lunch today? nelhage: I can ask anyone a question? [long pause, O(120sec)] tabbott: You are not obligated to ask the most devious question you can invent. nelhage: tabbott, a website with large quantities of javascript and CSS. We want to encourage the user's browser to cache aggressively so they are not downloading every request. We also want to make sure that if the content ever does change, we want to make sure they get the update immeidately, instead of caching the stale version. Give me an effective way to do this that has the property that as soon as we change the content on the server, it will update all clients, but clients don't have to make a request for it every time. tabbott: Hmm. I don't have a good solution for that. sipb3: The HTTP protocols have several headers including Expires, including an Apache module called mod_expires, or if you're using a scripting language you can solve those problems. nelhage: That's the traditional answer. But it's not so helpful, because if we set the expire time low we hose the server, and if we set it high clients may not see the updates. jhawk: I assume you could do something with AJAX. nelhage: Well, we can't keep a connection a connection open for days... nelhage: Our solution is to concatenate all your CSS together and name it according to the MD5sum. Assign the files a name based off of their content, so that every time a change happens, and clients get the new HTML that links to the changes. jhawk: How do the clients know when the HTML changes? nelhage: Part of the problem definition was that the HTML was frequently changing. jhawk: This is both a frivolous question, as well as a chance to show off your knowlege of MIT's administrative proceeses. How would you go about installing a firepole from the 5th floor to the 4th floor for the convenience of SIPB members who were also members of 4th floor activities. nelhage: I'd grab a hatchet and start chopping. aseering: I'd suggest a sawzall. mitchb: You could ask me and I would make some suggestions which would both work and piss jhawk off... mshaw: Why are analog phones sometimes better than digital?... When can a digital phone fail? nelhage: I believe that digital phones probably are required to be plugged into the wall? jhawk: Well, actually, they're powered from in-building battery backup that fails a few hours into a power outage. Whereas the analog phones are powered directly from the switch. mshaw: How long is rm? nelhage: "Forever." Except for OldFiles, so Not Necessarily. kchen: I move that Alex ask the question. [discussion of his uniqueness...] asedeno: So, Why do you want to be a SIPB member? nelhage: 'cuz I HATE the W20-5 cluster and the office has much nicer chairs. nelhage elected 10-0-0. tabbott: Congratulations, Nelson, you are now a SIPB member. MIT Computing Report: tabbott: The old MIT CA certificate expired today. mitchb: The Client CA, right? The regular cert expired weeks ago? jhawk: The Client CA *V1*. tabbott: I think that this caused script-certs to stop working jhawk: It also caused the Libraries Vera service to stop working and the controller's accounting web site to stop working, etc. Office Report: '() Office Cleaning Report: tabbott: What is this box? "Will be gone by mid-September"? What year is this September? SIPB Projects Report: mitchb: SIPB is attempting to run a cluedump series approximately weekly starting very shortly after computer tours in the hopes of a) providing people who have been interesting in topics that they don't know as much. ...to learn b) attracting frosh. mitchb: We have a handful of topics / starting points. We want volunteers to think of topics they'd like to teach or to teach items. Response has been sort of underwelming so far. Nelson doesn't suck. The rest of you suck. [jhawk makes hand gestures] mitchb: Oh, jhawk doesn't suck either, he's agreed to give a talk about SAP with me. jhawk: It oozes. mshaw: I'd be willing to talk if someone does most of the prep work. As in, if someone teaches me the material, I'm happy to form it into a presentation. mitchb: That's fine. Reply to the email with which ones interest you the most. mshaw: I only suck a little. tabbott: There will be a SIPB projects hackathon next Sunday, starting at 1:00pm, here. An hour before that, we will have an office cleaning that. You should make sure that your stuff that's like yours is in your drawer, as was announced previously. tabbott: Move to allocate $100 for the SIPB projects hackathon, the same amount as the previous one. Motion passes (mathmike seconds, jhawk white ballots) tabbott: The linerva project now has a reasonably stable setup on our testing virtual machine, and we are likely to reinstall our machine in the next week or so. The only thing that should apparently change is that we're moving to running only libraries from Debian Sarge, rather than the current odd mix which doesn't work all that well. So that weird problems when you try and build things don't happen. We also have these Debian-Athena packages that, if you would like to test, you should email us. We're tested them on Debian stable, and yoyo tested them on Ubuntu. Aside from the package depending on all libraries Athena has, which doesn't work for Ubuntu since they're missing some packages, things generally work. Let us know if you're interested. tabbott: Currently we have this list of available IP addresses. I think that all are in use by virtual machines somewhere. This is largely the result of scripts using 3 of them since last week. Investigating the Linux virtual server project, allowing you to take down production servers. I would like it to be true that when we run out of extra virtual IP adressess, I'd like to be able to un-deactivate some of our deactivated addresses without a motion. jhawk: How many are in use simultaneously? tabbott: Possibly all of them, as of right now. tabbott: Move to allocate up to $68/month for up to 4 static office IP addresses to be used to maintain the pool of spare office IP addreses. aseering: Second. Motion passes 6-0-1. tabbbott: We're also looking for someone to be the food sucker for Tours. This involves not going on Tours but getting food so it's there when people come back from tours. Please email sipb-tours@mit.edu and let us know, or volunteer right now. aseering: When exactly are the Computer Tours? tabbott: [discussion] Wednesday, September 6th, the day after Reg Day. jhawk: It's nice to have a car. tabbott: It'd probably be more nice to have a shopping cart. jhawk: Or a Hovercraft. tabbott: Or a Hovercraft! Publications Report: tabbott: I think some people are going to show up at the Hackathon with the intention to revise iAthena. So if you're interested in iAthena, please come to the Hackathon. CokeComm Report: '() Orientation Report: '() IAP Report: mshaw: Someone else say it. tabbott: IAP is sooner than you think. It's not a lie. mshaw: Would you like to be IAP sucker? tabbott: [to all] Would YOU like to be IAP sucker? tabbott: Normally we publish a reasonable number of AskSIPB columns during Orientation, modified versions of what we published previously, and then continue to run through the year. We did a pretty poor job this past year at continuing. If someone wants to take charge of the AskSIPB project for the fall, you should feel free to volunteer. yoz: If you have any questions about it you can ask me. tabbott: I believe there is one AskSIPB column in production, which is Katherine Lai working on something about Linux distributions. You should also feel free to produce a partial column. Other: '() Other Other: asedeno: Spam sucks. yoz: I set up a filter to fitler out all messages with sizes between 20k and 42k into a separate folder for inspection. So far all of those messages have been spam. mitchb: I'm going to be very careful about the mail I send to Yoyo from now on. mshaw: Harvard Computing sucks. Especially when my machine's memory breaks. tabbott: It just snapped? mshaw: No, it just died. The meeting was adjourned at 20:36:03. Minutes taken and submitted by jhawk.