Twitt-arrrr!

edited February 2012 in JoCo Cruise
Repost of the instructions, since the previous thread disappeared.

Note: Twitt-arrrr is now offline so that it can be transported down to FLL via secret Monkey Zeppelin.


Instructions on using Twitt-arrrr

It is not necessary to create an account or sign in to view Twitt-arrrr posts, but it is to be able to post.

Pre-boarding:
If possible, you should create an account right now on the Twitt-arrrr microblogging service.  To do so:
* Goto http://seamonkey.monster-island.org/statusnet
* In the upper right corner, click Login
* On the left hand side of the login page, click REGISTER
* Enter the nickname you wish to use as well as your password
IMPORTANT: For technical reasons this is not an encrypted site, so it is STRONGLY recommended that the password you enter here is one you currently do not use and that you will not use for any other website in the future.  It should not be something personal or important.  It should be something you wouldn't be afraid to tell a random stranger.
* All other information, such as email and real name are optional.  Put in as much or as little as you like, but don't enter websites or URLs that requires the Internet.  It might cause confusion.
* Make sure you accept the terms (last checkbox is ticked), and click the REGISTER button
* Feel free to explore and post messages

Post-boarding:
* Connect to the ship's WiFi network
* Note: you do NOT need to purchase Internet time to use the service.  It is internal to the ship.
* In your web browser, go to http://10.154.66.16/statusnet
* If you have already registered pre-boarding, simply click Login at the top of the screen and enter your username and password
* If you have not yet registered, in the upper right corner, click Login
* On the left hand side of the login page, click REGISTER
* Enter the nickname you wish to use as well as your password
IMPORTANT: For technical reasons this is not an encrypted site, so it is STRONGLY recommended that the password you enter here is one you currently do not use and that you will not use for any other website in the future.  It should not be something personal or important.  It should be something you wouldn't be afraid to tell a random stranger.
* All other information, such as email and real name are optional.  Put in as much or as little as you like, but don't enter websites or URLs that requires the Internet.  It might cause confusion.
* Make sure you accept the terms (last checkbox is ticked), and click the REGISTER button
* Feel free to explore and post messages

Twitt-arrrr has support for file uploads as attachments to messages and custom avatars.  Please limit your attachments to small to medium resolution images.  Twitt-arrrr is running on a wee little laptop and we don't want to overwhelm it.

Known Issues:
* For mobile devices like smart phones and tablets (iPhone, Android, iPads, etc), not all features work correctly.  The important stuff does, but things like uploading photos might not.
* There are a number of apps that claim to work with StatusNet (the thing Twitt-arrrr is made with).  They don't work anymore; this includes the official StatusNet app.  The website is the only way to use Twitt-arrrr, as far as we know.

Usage hints:
* There are a number of FAQs already written on Twitt-arrrr.  On the right hand side is the TRENDING TOPICS, and the faq topic should be one of them.  Click for a list of the FAQ answers.
* You do not need to follow someone to see their messages if you view the Public timeline.  Following is only needed for the sending and receiving of private messages, and for including their messages in your own Home timeline.
* Groups are a good way to send messages to a smaller number of SeaMonkeys on a particular topic.  If there is not a group already made for your topic, create one.
«13456716

Comments

  • The site for registering seems to be down right now.
  • Origami noted on the site that it would be down today while he travels to Fort Lauderdale.  Not sure when it will be back up.
  • Yes, it is down now. Sorry for the confusion. It is not going to be up until it is installed on the ship.
  • edited February 2012
    @Grimoire just make sure to watch out for submariners while you are on the secret zeppelin. Learn from the tragedy of Nick Mangold!
  • Actually someone else is transporting it down so it can be on the boat as early as possible for the installation. Godspeed Twitt-arrrr!
  • The Twitt-arrr! machine is now safely in FLL. :)
  • edited February 2012

    Just curious (and no biggie if not)-is it going to be up before it gets on the boat, or not until it's installed on the boat?  And I assume the public DNS name won't resolve to the internal IP since I doubt they're going to NAT it for us right?

    Unless we set up an internal DNS server...hmm doubt we're going that far eh? :P

  • Hey, we didn't think they'd even let us bring this thing along until this week. We decided not to add a rogue DNS server to the mix :)

    If everything goes as expected, the IP on the boat will be 10.154.66.16
  • edited February 2012
    And so it is. The problem, alas, is that the boat is using a Nomadix Wi-Fi gateway (built on Zyxel hardware). The maker of this gateway -- unlike my company, which also builds them -- charges the operator according to the maximum number of users the gateway will handle. Yes, there is literally a CHARGE to the company that operates the Wi-Fi to increase the size of the DHCP pool (the pool of IP addresses it can give out). And anyone carrying a Wi-Fi-enabled device can take up an address, whether he or she is using the Internet, Twit-arrrr, or nothing at all! As a result, service on the boat has been really shaky; I've had to apply a benign hack to get onto the network and do emergency work with my employees back home.

    The boat and its Wi-Fi operator are still obviously used to serving only technology-challenged snorks.

    If there's ever a land-based JoCoCon or w00tstock Festival, I will volunteer to do the Wi-Fi and make SURE it is top quality and reasonably priced. But I can't do much on the boat except put up with their flaky, spotty service. I'm writing this at a cybercafe on Aruba which is much more reasonably priced AND much faster.
  • @brettglass Yup, that pretty much sums up the issue. After talking to the IT officer at the end of the cruise, he was not at all surprised at the connection issues. He also mentioned that they are "upgrading" their system to double their IP pool. How MTN can justify such a system is beyond me. I almost expected to see a Gutenberg press in their server room, given their current config.
  • Indeed. @angelastic's strings and tin cans would have been an improvement.
  • Once I scanned the net and saw that addresses were being doled out in the 50 to 250 range... I started camping on a couple of under 50 addresses, with no adverse consequences.  I'll remember that for future cruises. :)
  • (De-stickying post-cruise.)
  • Thanks to everyone for helping make Twitt-arrr happen. We are sorry about the technical issues encountered; but suffice to say that MTN (the company that handles networking and wifi for HAL and numerous other cruise lines) are *very* interested in our results. And we have top men working on it already for next year.

    Top. Men.

    P.
  • @paulandstorm Excellent!

    I definitely got a lot of great feedback on the (hastily thrown together) Cruise Monkey app.  I'd love to extend it to integrate with twit-arr and such as well.

    I need to investigate Android's support for mDNS (Bonjour) but it might be an option to dual-home the Twit-Arr server on the automatic private IP network (169.x.x.x) as well as the HAL 10.x.x.x network, and let the app and browsers find the Twit-Arr server that way.

    I'll start a new thread on CruiseMonkey feedback, I've got other ideas I'd like to brainstorm as well.
  • Thanks for Twit-arrrr guys! It was awesome!
  • Twit-arrr (thanks for voting for the name I came up with) was a huge success, and helped ship-board communication greatly. Definitely looking forward to where this can go by the next cruise!

    Also, is it being put back online so we have the archive?
  • I'll try to get the DB to Scarface tonight. Last night involved too much unpacking and laundry and today involves too much work :-/
  • @paulandstorm - you guys are Top Men.  So noted.  :)

    To the rest - I think we have 12 months to give a small amount of grief to the iOS developers responsible for the StatusNet app.  Tell them to bring it current, and we pay them.  Very simple arrangement.  ;)

    Maybe.  The web interface was *okay*, but obviously iOS users could not post pictures.  The result - not many pictures got posted.  Not a knock against Android, but I saw a fair number of iPads, and personally my wife and I were carrying 4S'.  The DHCP problem was big, enough that I started suggesting we simply operate out of 169.254/16, but if our "Top Men" aforementioned are working out the kinks, they've earned my trust.  Superb job, everyone.

    Meanwhile, I'm making a note here to see if there's anything I can do to help beef up the hardware for next year.  I have a suspicion we're going to have a few more users next year.  This year was artificially limited to 200 concurrent.
  • @numbski The real issue with the app (Andriod and iOS both) is that the web developers KEEP CHANGING THE API! Every dot release broke the apps. I blame whoever is working on the API for not communicating and for making it a moving target.

    Also, talk to me about the 169.254 IP range, as networking is not my strong suit. Could we connect people using just 169 and Bonjour services? What about people who CAN get a 10. address via DHCP, would they be blocked out? What services could we reasonably run that way? /devious
  • edited February 2012
    I *think* if the server can be dual-homed (on both the 10.* network and the 169.* network) it should work OK regardless of how you end up being connected.  I'll see if I can do some VM tests to see if resolution works as expected.  The fun part is to dual-home, we'll probably have to do a script or something to do arp tricks and get a free 169 address, since at least on my system, it won't negotiate a 169 address unless it can't get a normal one.

    Anyways, in theory, the mDNS responder should listen on all interfaces and if someone asks for the IP for "Twitt-arrrr.local" it will respond.  In practice, I'll do a little experimentation to confirm.  :)

    At that point, you can run whatever services you want, and it should be happy.
  • Okay, so in short, here's what I'm thinking (presuming you're on linux and presuming we're back on the Westerdam).  They gave you a static 10.x address, right?

    So that means ifconfig eth0 (or 1, depending0 should have had 10.x.x.x assigned to it, with a netmask 255.255.255.0, with matching netmask and gateway.  The other devices on board were in that same netmask.  If you're getting DHCP assigned addresses, and you fail to get a lease, you'll get 169.254.x.x with a netmask of 255.255.0.0 and no gateway.

    So I say you add a virtual interface to the box, call it eth0:0, and assign it whatever 169.254.x.x address contents you, with netmask 255.255.0.0.  Go ahead and leave eth0 as it was.

    What then happens is you configure statusnet to answer on any available IP, be it localhost, 10.x or 169.254.x.  It was probably that way by default.

    The end result is that we have 2 available Twitt-arr addresses.  If you got a lease, it's 10.x.  If not, 169.254.x.x.

    Could we advertise using Bonjour?  Absolutely.  Say we assign the box the name twitt-arrr.local.  We set up Avahi to advertise the correct IP on each separate interface.  That way no one has to remember what address to use:  twitt-arrr.local will always answer.
  • And users will auto-magically self assign a unique 169 address? Are there limitations as to what will respond to, say, seamonkey.local (to allow for more than just twitt-arrr. I think a calendar server was already mentioned)?
  • Yup.  If they get a 10.x address, great.  If not, they'll get a self-assigned.  With multicast dns, it should be totally seamless.

    Of course, we'd want to test that in advance...
  • Speaking of testing, I never got around to testing Twit-arrr without internet access. I guess it works :)
  • Are the 169 addresses really unique? My understaning is they self assigned when dhcp fails i.e. there is potential for ip conflicts. I have no idea what the various algorithms are for assigning the address are; and among different OSs they might be different again, hopefully 2 octects is fairly safe. I suppose if it were a declared limitation the more savey users could just hunt for their own addresses.
  • They will be unique -- the client does a gratuitous ARP before grabbing an address to make sure it's unoccupied.
  • Gratuitous ARP is the name of my Daft Punk cover band.
  • LOL, yeah.  This is the point where we flex our nerd muscles and link to RFCs, right?

    I will definitely do some testing with various combinations of DHCP availability and non to make sure everything works, but I expect it should do the right thing either way, as long as we manually make the twit-arrr server live on both subnets.
  • Thanks to Team Twit-Arrrr for some fabulous work. It was awesome enough that I understand both MTN and Carnival Corp had people paying close attention to it.
  • edited February 2012
    I've arrived home and had a day or so to unwind from the traveling.

    In addition to the Bonjour name service, it would probably be a good idea to run a NetBIOS Name Service via Samba.

    I wouldn't mind reading the RFC that discusses the 169.254 address allocation.  I'd be interested in reading up on how it should work if you happen to know the RFC number.  If not, I can probably google it.  :)
    ETA: RFC-3927

    One thing I noticed that might be a problem was that when my iOS devices did not get an IP address via DHCP, they would disconnect from the wireless network.  I don't know if this was a server or client side disconnect, or if it is configurable.
  • I just skimmed the DHCP RFC.  I think what was happening was the DHCP server was responding to DHCPDISCOVER requests, but since the available IP addresses was exhausted, it was not responding to DHCPREQUEST requests or was always responding with DHCPNAKs.  Since the device saw a DHCP server but couldn't get an address, instead of auto-assigning an IP, it would drop the connection.

    In hindsight, I really wish I had run a packet capture of the failed connection attempts.

    In the DHCP RFC, I didn't see what the server should do in case of complete IP exhaustion.  If the DHCP server stops responding to DHCPDISCOVER requests, the device might stay connected to the network and then auto-assign an IP address in the 169.254/16 range.  I don't if that is compliant behaviour, nor if MTN would be able to reconfigure/recode their servers to act accordingly.
  • edited February 2012
    Alas, MTN cannot recode or reconfigure its servers to get around the limitations, because the vendor from whom it buys them -- Nomadix -- imposes arbitrary limits and uses them to extract more money from the purchaser. (Want more IP addresses in the DHCP pool? Pay more for the hotspot server.)

    The best way to get around the limitations without causing MTN or HAL any grief or expense would involve setting up some of our own wireless routers in key parts of the ship, such as the function spaces, Lido Deck, and poop deck. They'd be bridged to one another through the ship's network but wouldn't take up any IP addresses in the ship's subnet at all, so limits on the ship's DHCP server wouldn't be an issue. (In fact, we'd unload the ship's hotspot by removing Sea Monkeys' devices from the competition for addresses.) Giving the routers a common WPA2 password (which we would announce and post in the game room) would keep snorks from connecting to them by mistake. The result would be a Twitt-arrrchitecture that was completely independent of the ship's hotspot server and used only a little of their Wi-Fi capacity.
  • Ooooh, I like that idea a lot.  If you want to use your "real" internet minutes, connect to the MTN network, if you want MonkeyNet, connect to it instead.  No confusion, and better control.  (At the expense of ubiquity, but in practice, going to one of the common SeaMonkey places to "sync up" would be more than enough.  Also, if it's an inTRAnet with no inTERnet, perhaps it would satisfy the WiFi Temperance Brigade concerns.)
  • I suspect David Rees pushed to bar Twitt-arrr just for effect (especially since his explanation involved beating the event schedule out of a staff member) but I expect there will be people who want to unplug for the duration of the trip. 

    The biggest issue I can see to MonkeyNet is where to put the routers where they will function but be out of site. Also HAL might not be fond of us setting up a shadow network on their boat.
  • edited February 2012
    It looked like there were plenty of places to stash them in public areas -- e.g. behind bar counters. And, of course, the game room and auditorium would be no problem; they could be in plain sight.

    As for setting up a "shadow network:" that's really what we did this time. We'd just be making ours more separate and cause it to use fewer of their resources -- two things I cannot imagine they wouldn't approve. And if we used 1 watt access points, the signals would reach to a lot of the cabins.
  • Personally, I found Twitt-arrrr most useful outside of the function spaces. I used it most in my stateroom and while wandering the ship looking for unofficial events. I think we would see a lot less usage if we didn't have significant coverage of the ship.

    The HAL IT officer did mentioned that they were doubling their IP range, and that many other ships were doing the same. That, combined with improved WiFi etiquette (disabling WiFi when not in use) would greatly improve the situation to acceptable levels.
  • I started disabling my wifi when not in use, and found that when i re-enabled it I couldn't get a connection again.
  • @ReptilianSamurai Most likely because not enough people disabled wifi when not in use.  ;)
  • Game theory rears its head.
  • The obvious solution is for everyone to wear a wifi repeater on their persons at all times, so we can make a mesh network.
  • Just don't wear it at the airport, especially if you're flying out of BOS.  The cops here have a reputation for being incredibly skittish about electronics.
  • WiFi repeater fezzes with stachetennae.
  • edited March 2012
    [Restrains self from posting obvious pun regarding Sea Monkeys making a mesh]
  • @mtgordon - I too am sad about our town's anti-electronic paranoia.  You'd think that with such a strong innovation economy here people would be excited to see DIY kits instead of irrationally afraid.
Sign In or Register to comment.