Princeton University Details iPad Wireless Networking Issues

[Iphtashu Fitz]

Well the last time something like this came up with iPhones at Duke the blame was put on Apple when it was later discovered to be an issue with the Cisco equipment. If it really is the iPad then why is it an isolated case?

Most likely because of the nature of Princeton's network setup. I work at a different university, and our DHCP environment is set up in such a way that the DHCP servers remember what device an IP address was assigned to, and will reuse the same IP as much as possible. Many DHCP servers do this. So even if a client doesn't renew a lease properly and continues to use the IP it won't necessarily cause a problem. Think of it like the DHCP server extending a reservation implicitly for some number of hours.

It sounds to me like Princeton not only has a short DHCP lease time but it also cycles through IP's rather quickly. So as soon as an IP is considered to be expired it quickly is re-assigned to another device. It's the combination of the device using the IP beyond the defined lease time AND the DHCP server re-assigning the IP to a new device that causes this problem. If the DHCP server always used the least-recently-used IP for a new reservation and they had a good sized pool of IP's then this likely wouldn't have happened.

 

[alent1234]

REad a more extensive article last night and it is more Princeton than what has been reported by other Universities.

You know the drill, IT doesn't want Apple hardware on their systems and are unwilling to trouble shoot issues with the network.

in this case it's up to the client to release the IP. the network cannot make a client that has a OS bug not answer the wifi equivalent of an ARP request

 

[schmidt349]

Yet another Princeton student here (Ph.D), and I'll second everything that's been said about the Princeton network situation. OIT is incredibly strict about the behavior of devices on the wireless network, and this includes much more than just DHCP misconfiguration. Even the slightest instance of nonstandard device behavior is punished with an immediate ban from the network. Moreover, devices exhibiting "nuisance" behavior can also be banned, usually with a letter tinged with moral opprobrium about how horrible a human being you are because you accidentally left Network Sharing turned on when you reconnected to the campus network.

What makes their behavior all the more mystifying is the fact that the wireless network here is a complete shambles. They use one subnet for all wireless traffic everywhere on campus. This results in your device being flooded with about 10k/sec of trash traffic like ARP requests and other stuff that should be subnetted and routed so that you never see it. If you have a Unixy device with a badly configured ipfw it will send your CPU use through the roof. I worked in IT at the University of Chicago for four years as an undergrad so I've seen what a far saner wireless environment can look like.

If the iPad is not doing DHCP correctly then Apple needs to fix it pronto, but Princeton's draconian and grossly inefficient network policy does not facilitate matters at all.

 

[alent1234]

Most likely because of the nature of Princeton's network setup. I work at a different university, and our DHCP environment is set up in such a way that the DHCP servers remember what device an IP address was assigned to, and will reuse the same IP as much as possible. Many DHCP servers do this. So even if a client doesn't renew a lease properly and continues to use the IP it won't necessarily cause a problem. Think of it like the DHCP server extending a reservation implicitly for some number of hours.

It sounds to me like Princeton not only has a short DHCP lease time but it also cycles through IP's rather quickly. So as soon as an IP is considered to be expired it quickly is re-assigned to another device. It's the combination of the device using the IP beyond the defined lease time AND the DHCP server re-assigning the IP to a new device that causes this problem. If the DHCP server always used the least-recently-used IP for a new reservation and they had a good sized pool of IP's then this likely wouldn't have happened.

this depends on the lease time. if the lease is 5 days or so than it's not a big deal. a lot of organizations like a short lease of 1 day. makes it easy to prepare for downtime. we used to change it to 4 hours if we had to do something that would result in IP conflicts when we were donw

 

[MacCheetah3]

Hi

Well the last time something like this came up with iPhones at Duke the blame was put on Apple when it was later discovered to be an issue with the Cisco equipment. If it really is the iPad then why is it an isolated case?

I'm not saying it isn't a 'bug' with the iPad software but I seem skeptical as well. I know the router(s) I used at work had a few moments of assigning the same IP to different devices ( clients ) and needing to be reset. It's annoying and not common but seems to be more of a router mis-assignment than the client device(s). Unlike most residential uses of routers, where the devices are 'static' for an extremely long time, my use and large organizations like schools have a lot of different devices constantly connecting and disconnecting and it just seems that sometimes routers get confused. Since my business was computer repair, I had the same type of situation.

 

[*LTD*]

Magic?

I'd love to see what people have to say about the "other" slates and tablets out there. Not a pretty situation.

They make iPad issues look like jellybeans.

 

[Rocketman]

There were handshaking issues with the AT&T wireless network. I wonder if these issues are related?

Rocketman

 

[mdatwood]

What bothers me is that this is basic DHCP client functionality. Just another example on the decline of Apple QA

 

[Iphtashu Fitz]

I'm no wireless expert, but I found OIT to be really draconian with their network. I ran into several problems with them with my iPhone, trying to use a time capsule, and setting up wireless access for a Wii. I really don't know enough to be sure, but I am at least suspicious as to whether these iPads are really causing problems vs. just not behaving "properly".

The university I work at puts all sorts of restrictions on the network ports in residential halls, including not allowing use of wireless access points like Time Capsules. If more than one MAC address is found accessing a port (a sign that a switch or wireless hub has been added) then the port will be shut down until the device is removed.

There's a few reasons for them doing this. First, our university already provides wireless access so there's no need to have your own in most cases. Second, insecure wireless access points can be a security concern. We've had cases of spammers using open or poorly secured wireless devices to pump spam through our networks.

 

[jragosta]

i'm shocked Apple QA didn't see this in their super secret locked down lab with only one person authorized to touch the iPad

And your evidence that Apple didn't do QC? It's really amazing how many people jump on "Apple should have caught this" without knowing the specifics. In most of the world, this would be a difficult thing to catch. It applies at a University because you have thousands of users all using the same access points at the same time - and staying connected for many hours. That's not likely to crop up in real life.

Apple's QA completely sucks. They're a company that wants everything to be wireless, and their devices don't even perform properly on a basic level.

And, yet, Apple customer satisfaction and reliability is the top of the heap. Sounds like their QA is fine - you just have unreasonable expectations ('no problems, ever, no matter how a product is used).

Well the last time something like this came up with iPhones at Duke the blame was put on Apple when it was later discovered to be an issue with the Cisco equipment. If it really is the iPad then why is it an isolated case?

Exactly. While I don't doubt that the iPad's use of DHCP is a problem, it's also a rare problem - which is why it wasn't caught. It also suggests that Princeton is doing something that other sites don't do, contributing to the problem.

Speaking as a student currently enrolled at Princeton, I do want to note that Princeton is immensely anal/aware of networking snafus - they are often quite willing to block students who may have accidentally messed the network up somehow..

Princeton is clearly tracking this issue very carefully and has developed solid diagnostic information. That will help resolve the problem - no matter who is at fault.

Good call. We had a similar issue. We escalated to our Apple rep who told us our router was configured wrong

And did fixing the router fix the problem? 95% of the time, it does.

For long document creation, you'll wish you had a physical keyboard, trust me.

What's your point?

The iPad is primarily a content consumption platform, not content creation. If you want to do heavy duty content creation on an iPad, you can use an external keyboard.

Your comment is akin to saying that Corvettes are bad because if you want to haul half a ton of top soil, you'll want a pickup truck.

 

[bludodge]

See, My issues were not my imagination. This is why I took my unit back. I miss it dearly. Maybe I will get one again when they work this all out.

 

[baryon]

I have absolutely no idea what these complicated words mean.

 

[min_t]

Update fix will be in OS 4. No need to rush. It's not like everyone has an iPad.

 

[-=XX=-Nephilim]

It must be Adobe's fault!

>_<

 

[Evora]

I finished my PhD at Princeton a couple of years ago, and I just thought I'd chime in to agree with what others have said here. Princeton IT was generally passive agressive and totally inflexible. I've never known a more unreasonable IT department. There may be an issue with the I-pad but it's telling that IT departments at other institutions (including the one where I now teach) haven't had this issue or have found a work-around.

 

[Hal Itosis]

Yet another Princeton student here (Ph.D), and I'll second everything that's been said about the Princeton network situation. OIT is incredibly strict about the behavior of devices on the wireless network, and this includes much more than just DHCP misconfiguration. Even the slightest instance of nonstandard device behavior is punished with an immediate ban from the network. Moreover, devices exhibiting "nuisance" behavior can also be banned, usually with a letter tinged with moral opprobrium about how horrible a human being you are because you accidentally left Network Sharing turned on when you reconnected to the campus network.

What makes their behavior all the more mystifying is the fact that the wireless network here is a complete shambles. They use one subnet for all wireless traffic everywhere on campus. This results in your device being flooded with about 10k/sec of trash traffic like ARP requests and other stuff that should be subnetted and routed so that you never see it. If you have a Unixy device with a badly configured ipfw it will send your CPU use through the roof. I worked in IT at the University of Chicago for four years as an undergrad so I've seen what a far saner wireless environment can look like.

If the iPad is not doing DHCP correctly then Apple needs to fix it pronto, but Princeton's draconian and grossly inefficient network policy does not facilitate matters at all.

Nice try, but good posts like yours seldom get the attention they deserve.

[thanks... now let the unbounded criticism of a product barely 3 weeks old continue.]

 

[locust76]

iPad:

it can't multitask
it can't run a real OS
it can't take photos
it can't communicate properly over a network

 

[trekkie604]

Wirelessly posted (Mozilla/5.0 (iPod; U; CPU iPhone OS 3_1_3 like Mac OS X; en-us) AppleWebKit/528.18 (KHTML, like Gecko) Version/4.0 Mobile/7E18 Safari/528.16)

See, My issues were not my imagination. This is why I took my unit back. I miss it dearly. Maybe I will get one again when they work this all out.

If you just use it at home it's easy enough to work around. Just give the ipad a static ip and avoid the dhcp issue until it's resolved.

 

[*LTD*]

iPad:

it can't multitask
it can't run a real OS
it can't take photos
it can't communicate properly over a network

You have a very narrow view about how people should interact with and approach technology.

 

[locust76]

Oh btw, this isn't Apple's fault. Apple is simply using DCHP5, the new, open DHCP standard. It's the industry that needs to catch up!

 

[alent1234]

And your evidence that Apple didn't do QC? It's really amazing how many people jump on "Apple should have caught this" without knowing the specifics. In most of the world, this would be a difficult thing to catch. It applies at a University because you have thousands of users all using the same access points at the same time - and staying connected for many hours. That's not likely to crop up in real life.




And, yet, Apple customer satisfaction and reliability is the top of the heap. Sounds like their QA is fine - you just have unreasonable expectations ('no problems, ever, no matter how a product is used).



Exactly. While I don't doubt that the iPad's use of DHCP is a problem, it's also a rare problem - which is why it wasn't caught. It also suggests that Princeton is doing something that other sites don't do, contributing to the problem.



Princeton is clearly tracking this issue very carefully and has developed solid diagnostic information. That will help resolve the problem - no matter who is at fault.



And did fixing the router fix the problem? 95% of the time, it does.



What's your point?

The iPad is primarily a content consumption platform, not content creation. If you want to do heavy duty content creation on an iPad, you can use an external keyboard.

Your comment is akin to saying that Corvettes are bad because if you want to haul half a ton of top soil, you'll want a pickup truck.

DHCP has been around for decades. in my 15 years or so of playing and supporting computers i've never seen a client device of any OS with this bug

 

[mikethebigo]

What's your point?

The iPad is primarily a content consumption platform, not content creation. If you want to do heavy duty content creation on an iPad, you can use an external keyboard.

Your comment is akin to saying that Corvettes are bad because if you want to haul half a ton of top soil, you'll want a pickup truck.

Dude, what's with all the hate around here? I didn't say the iPad was a bad device, I own one for christ's sake. I know its good for consumption, i never said it wasn't.

The guy above my comment SPECIFICALLY asked about long content creation, and i told him he'd want a physical keyboard. You, apparently, agree with me on that point. Whats the big deal?

 

[FlagellumDei]

Your right!! I completely forgot!! it should be

if ( [lease isExpired] && [SteveJobs saysYes] && [iPhone4g isSecret])
[device requestLease];
else
[iPhone4G createLeak];
lol

That was a nice one!

 

[macUser2007]

Oh btw, this isn't Apple's fault. Apple is simply using DCHP5, the new, open DHCP standard. It's the industry that needs to catch up!

LOL! Princeton is just lazy...!

Macs had a similar problem (expecting the same IP address forever), if my memory serves me right.

 

[Mabus51]

That's what I was afraid of. At that point, it seems like for the cost I might as well get a macbook or even a netbook for much less for quick boot, note taking and document creation.

If you really want a physical keyboard they make one for the iPad. I'd rather have a handwriting tool so I could scribble my notes on the iPad more so than typing them out. Just me though