computers

UTOSC 2009 Is Full Steam Ahead!


And with that, I've uploaded all my slides for the Utah Open Source Conference. My goal was to have them ready by last night, so I'm a bit late. Still better than last year where I only had one presentation and I was still updating my notes the night before. If you're looking for me make sure to check the schedule as there have been some changes, and there may well be more to come. Also look for my late addition where I'll be joining Gabe Gunderson of izeni to talk about VoIP (specifically FreeSWITCH and Asterisk).

tags: 

Presenting at UTOSC 2009


I will be presenting at the Utah Open Source Conference (UTOSC) again this year. On tap are 3, count 'em, 3 presentations. I'm still not quite sure what possessed me to submit 3 abstracts but I did and (just to spite me I think) they accepted all of them.

Refer to the schedule for any last minute changes but for now here are the times I'm speaking.

Hope to see you there.

tags: 

Mail Client Duldrums

Lately I've been feeling rather disappointed by Evolution. I've been a long time user (7 years I think) and for the most part it works great. But it's that "most part" bit that really is starting to grind. The last few iterations that I've tried (whatever is bundled with Ubuntu Feisty, Gutsy, Intrepid and Jaunty) have all had a a few quirks, none of them the same of course, For instance, The Intrepid version had an annoying habit of leaving messages marked as unread, even after I, you know, read them. Made for an annoyance when filing messages away. Worse, many of the messages that Evolution said were read, weren't really marked as such on the server so sometimes messages would magically unread themselves.

So I decided to give KMail another shot. I tried it last year sometime I think it was and decided it just didn't cut the mustard. But I'm a few revs forward on KDE now so it was worth a try. Over the last few days I have found it to be less annoying and much snappier than Evolution so I am considering a permanent switch, but it too does have issues. One major oddity is that when I open up a new folder, the unread messages count will reset while the folder is rescanned. That's just crazy. I can see why in a way, but there's just no need for it. A second complaint is that there's no way to move to the next message without closing the currently open message and opening a new window (I turn off the preview pane).

On the other hand, KMail excels in a quite a few ways. Contact auto completion is much much (much!) faster. I like the idea of the "favorite folders", although to date I haven't made much use of it. I like the way it integrates with my Spam Assassin. It handles multiple identities perfectly (although the configuration is kind of spread out).

Of course the elephant in the room is Exchange support. It's a necessary evil at least at my company (better than Groupwise!). Evolution has an Exchange plugin which works via Outlook Web Access to give you all the features of Exchange in Linux. In theory anyway. While many features do work, not all are flawless. The address book doesn't work for me. My calendar doesn't seem to sync with the server calendar. I've had previous problems with server-side calendar reminders not working, although the version in Jaunty seems to work fine.

Is there anything better out there that I should take a look at? Exchange support would be great, but frankly I don't use all the extra features all that often and if I have to fire up Outlook (via terminal services) once a week, that's not a deal breaker. I would like it to be rock solid, though.

tags: 

tcpreplay

I'm working on an upgrade of a RADIUS server and I need the ability to verify that my changes won't alter the behavior of the server. So what would be really nice is a way to record all the network traffic going to my RADIUS server with a tool like tcpdump and then resend it to my test server and compare the results. As luck would have it, there is such a tool and it's named tcpreplay.

While I haven't run the full tests on my RADIUS server, I have done a few simple tests with ICMP and UDP packets just to verify that it will work with protocols other than TCP, despite its name. It does. Here's an example.

In on root shell, run the following command to capture packets:

# tcpdump -np -s0 -i eth1 -w icmp.pcap icmp and dst host 192.0.20.1

Then, in another shell, start a ping to the IP address in question:

$ ping -c 5 192.0.20.1
PING 192.0.20.1 (192.0.20.1) 56(84) bytes of data.
64 bytes from 192.0.20.1: icmp_seq=2 ttl=64 time=1.63 ms
64 bytes from 192.0.20.1: icmp_seq=3 ttl=64 time=1.49 ms
64 bytes from 192.0.20.1: icmp_seq=4 ttl=64 time=1.55 ms
64 bytes from 192.0.20.1: icmp_seq=5 ttl=64 time=1.55 ms

--- 192.0.20.1 ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4021ms
rtt min/avg/max/mdev = 1.490/1.567/1.639/0.049 ms

Now you've got a PCAP file that you can feed to tcprelay. This is a very basic, and fun, way to run tcprelay so that you can watch and confirm each packet. There are many other options for how you can alter the replay.

# tcpreplay --intf1=eth1 --oneatatime --verbose icmp.pcap
sending out eth1
processing file: icmp.pcap
reading from file -, link-type EN10MB (Ethernet)
15:45:37.376377 IP 192.0.20.189 > 192.0.20.1: ICMP echo request, id 58216, seq 1, length 64
**** Next packet #1 out eth1. How many packets do you wish to send? 1
Sending packet 1 out: eth1
15:45:38.383298 IP 192.0.20.189 > 192.0.20.1: ICMP echo request, id 58216, seq 2, length 64
**** Next packet #2 out eth1. How many packets do you wish to send? 1
Sending packet 2 out: eth1
15:45:39.391925 IP 192.0.20.189 > 192.0.20.1: ICMP echo request, id 58216, seq 3, length 64
**** Next packet #3 out eth1. How many packets do you wish to send? 1
Sending packet 3 out: eth1
15:45:40.394081 IP 192.0.20.189 > 192.0.20.1: ICMP echo request, id 58216, seq 4, length 64
**** Next packet #4 out eth1. How many packets do you wish to send? 1
Sending packet 4 out: eth1
15:45:41.398076 IP 192.0.20.189 > 192.0.20.1: ICMP echo request, id 58216, seq 5, length 64
**** Next packet #5 out eth1. How many packets do you wish to send? 1
Sending packet 5 out: eth1
Actual: 5 packets (490 bytes) sent in 15.14 seconds
Rated: 32.4 bps, 0.00 Mbps/sec, 0.33 pps

Statistics for network device: eth1
Attempted packets: 5
Successful packets: 5
Failed packets: 0
Retried packets (ENOBUFS): 0
Retried packets (EAGAIN): 0

tags: 

New Phone: Samsung SCH-i760

My boss surprised me the other day with a new phone. Nothing was particularly wrong with my old phone, but we had a bunch of these new Samsung phones due to some staffing changes so I got nominated for one. Ostensibly it's to make on-call easier since I'll be able to read and respond to emails without having to run home. Hopefully that will work out.

So far I think the phone will work out alright. It's fairly bulky, but so was my LG enV, so no big change there. The screen seems big enough to be functional, although it certainly has its limitations. Can't say I can see much solution there short of packing around a netbook though. I love finally being able to synchronize my calendar, something that in theory the enV can do but which Verizon (curse their souls) forcibly disabled. Mobile email also cool. I've installed some applications such as Facebook, Twitter, IRC, that may or may not be really useful in the long run but are fun to have around.

But of course there's bad news too. The battery life is not what I'd like it to be. I'm not sure how it compares to other smart phones since this is my first, but with any reasonably heavy usage I'm looking at charging it a few times a day, and that's going to be my usage pattern during on-call, so I'm a bit disappointed.

It also runs Windows Mobile. I was rather apprehensive about it, but so far it's doing better than I thought. I'd give it a solid mediocre (which is high praise from me).

The application support is disappointing, although not really unexpected. See, with a Free/Open Source background I expect high quality software to just be available for free. With Windows Mobile, yeah not so much. There are a few but they're somewhat hard to find. I could really use an apt-get if you know what I mean. This is an area where Linux users get pampered and it's tough to go without.

Those are the major drawbacks. The one other minor glitch I would mention is that whenever I'm connected to my wifi it constantly tries to bring up a data connection. Well what's the point of that? To get the thing (I named it Aristotle) to knock it off I set up a dummy ISP connection using the irDA port. Hee, hee.

It's a keeper so far. We'll see how I feel after I go on-call this weekend. If you see this blog full of Brittany Spears posts you'll know it's driven me off the deep end. Your pity is all I ask.

tags: 

A Rose By Any Other Name

Stephen Dubner, the famed economist behind Freakonomics (an excellent book, btw), posed the question, is it time to rename 'Digital Piracy'?" The answer is an unequivocal "yes". Despite what Shakespeare said, sometimes a name can mean everything. I refuse to accept the word "piracy" as anything other than high-seas pillaging. I suppose one positive outcome of the recent surge in Somali piracy is that people are realizing that copyright infringement hardly warrants such a strong word, especially when we have a perfectly adequate one. Let's not let the geniouses at the RIAA, who've had a wonderfully successful program of suing their customers, dictate our terminology on the matter.

tags: 

Password Security

Great article here about poorly chosen passwords. I tend to stay away from simply relinking URLs that I find, but I think this one merits some attention. I recall when I was studying my tail off in college trying to learn everything about computers, especially security, that passwords were a rather common topic. The recommendation then was to use as a password difficult enough to not be guessed but still simple enough to be remembered. I'm pretty sure that advice fell on deaf ears as frankly it's pretty tough to do.

Any more I recommend using as long a string of random characters as possible. Mash on your keyboard if you have to, but also consider a random password generator. How will I remember that, you might ask. Don't. Use a password safe. Then all you have to do is remember one password. The rest are kept in a vault for easy access. And if you can't remember your master password, write it down and keep it in your wallet. Somebody steals that, well you've got bigger problems.

Once you adopt this new technique, you'll use passwords differently. Instead of reusing the same one, possibly with variations, you'll generate a completely different one for every website. Remember the old saying about keeping all your eggs in one basket? Once in a while at work we'll have accounts compromised because the user had the exact same password everywhere and some spammer nabbed it and then hijacked their email account to send spam. Sounds crazy, but you definitely do not want that to be you.

And while you're at it, stop using 6 letter passwords. It's all random and you'll just copy and paste it anyway, so why not go for broke? I routinely use 24 character passwords. It actually makes me smile to think of how ridiculous my password would seem to a snooper.

tags: 

Asterisk - Call Progress And Early Media

When you make a phone call, say to your grandma, you hear her phone ringing. It's called ringback and its purpose is fairly obvious. But have you ever wondered where it comes from? As a kid I assumed I was hearing grandma's phone ringing which, if you think about long enough, is kinda ludicrous. I haven't read up enough to know where it actually came from but I can tell you where it comes from in a SIP environment. There are basically two options.

First of all, the called party sends a 183 Ringing message to the caller. That's how your phone knows that grandma's is ringing. Your phone also then plays some sort of ring noise to let you know.

There's another possibility though. Instead of just saying 183 and being done, the callee can send back an RTP stream with the ringing noise of its choosing. I'm sure you've heard different types of ringback before (different pitch, different frequency, different volume) and that's because voice switches each have slight variations on the same theme.

Now, take that one step further and consider that an RTP stream is just an RTP stream and doesn't have to conform to some traditional ringing sound. It could be beeping, or honking, or whatever you want. Screeching monkeys, did I hear you say? Oh yeah. But probably more useful is the ability to send music. Here is how to set it up in Asterisk. (BTW, this will work with SIP, IAX2, or PRI. Basically any of the cool signaling methods, but it won't work with analog POTS.)

First we set up a music on hold class which is our ring noise. This can contain any file you want, but since Asterisk will play it from the beginning each time, best to be something that sounds cool immediately. Or pull it up in Audacity and trim out your favorite 60 seconds.

musiconhold.conf:

[ring1]
mode=files
directory=/var/lib/asterisk/mohmp3/ring1
random=yes

Then in our dialplan we use it. First we play a message to the caller that we're trying to find the user, otherwise they might get confused as to why they are hearing music. Then we pass the "m" option to Dial which specifies our music on hold class. It's really as easy as that.

extensions.conf:

[from-pstn]
exten => s,1,Progress
exten => s,n,Playback(followme/pls-hold-while-try)
exten => s,n,Dial(SIP/you||m(ring1))

tags: 

Ma Bell Innovates?

It appears that Qwest is adding some actual features to their phone service. The new service "allows users to see who’s calling their home phone, listen to voicemail, forward messages, manage their contacts and place a phone call." Actually sounds pretty cool. It's nice to see that a 800lb gorilla out there isn't just resting on its laurels. Still isn't enough to convince me to switch back though.

tags: 

Scamware

CNet is reporting that there is a crackdown on so called "scamware", software that tells you that something is wrong with your computer and thus you need to buy some software which they conveniently will sell to you. I think my wife ran into some of that today. She was looking for info on how to fix the rear wiper of our van (pop the cover and tighten the 10mm nut, if you were wondering). One of the sites popped up another that said she had some Win32 virus and she needed to install their anti-virus software. Well, since she was on an Ubuntu machine there was exactly 0% chance that she really had the virus, so something was definitely awry. No clue if that particular software is from the vendors that Washington state is targeting, but it sure was interesting to run across the story on the same day.

tags: 

Pages

Subscribe to RSS - computers Subscribe to zmonkey.org - All comments