Jump to content

Wikipedia:Reference desk/Archives/Computing/2010 August 10

From Wikipedia, the free encyclopedia
Computing desk
< August 9 << Jul | August | Sep >> August 11 >
Welcome to the Wikipedia Computing Reference Desk Archives
The page you are currently viewing is an archive page. While you can leave answers for any questions shown below, please ask new questions on one of the current reference desk pages.


August 10

[edit]

LAN authentication

[edit]

[Preambulatory warning: As an end-luser of a little computer (not a programmer or system administrator or anything), I'd rate my knowledge and abilities B minus or C plus; but as somebody wanting to connect it up to the wider world, I'd rate them C minus.]

In the large institution in which I work, ethernet jacks and wireless are plentiful. One office where I often work has a number of desktop computers I can use, each connected by wire. When I turn on any of these computers, open a browser, and ask it to look at any page, I'm instead given my institution's standard https page in which I type my ID and password. (Let's call this the login page.) Once I've done this, I can websurf freely. (No bittorrents though.) Perhaps I ought to be able websurf endlessly, but in practice I can't: One computer (A) gives me the login page after ten minutes or so, another (B) after twenty-five minutes or so, the third (C) after two hours or so. I haven't actually timed these, but all of us who use them agree on our impressions. It's particularly odd as A and B have been plugged into the very same router -- or hub? anyway, little doodad with tiny flashing diodes -- and have been like this since the day they were brand new and set up with identical (Windows) OS and software.

I didn't much care about this matter until yesterday, because until yesterday my most-used computer, in my own little office, never had the LAN again give me the login page, unless the computer had been awoken from sleep (or of course had been turned off and back on or had had the cable temporarily removed and plugged back in). However, yesterday -- but no, let me pause at this point for now.

Why might computers A and B behave differently, and how might I increase the time for A? If I know the fix for this, I might be able to fix the brand new, serious irritation within my own office. -- Hoary (talk) 02:41, 10 August 2010 (UTC)[reply]

This rarely is a setting on the computer itself. It is a setting in the institution's DNS resolution. Initially, no matter what you type, be it google.com or iwanttoseebigboobies.com, the IP address that gets resolved will be the institution's local login page. After you properly login, the computer's MAC address is saved for a preset time in the DNS to allow it to properly resolve addresses. Then, you can surf the web all you like. Eventually, the MAC address will be kicked off and you will have to login again. Understanding that, you can see that there is a way around it. You can type in an IP address instead of doing DNS lookup. For that matter, you can use an alternate DNS address. But, this entire thing is usually built into the institution's switch system so you can't get around it like that. If the computer isn't authenticated, it won't be allowed to talk to anyone except the login machine.
Now, after saying all of that, it is possible that the institution is doing something rather silly - putting the authentication on the machines themselves. If so, it is easy to bypass. Just use a live disk and avoid using the computer's actual operating system. Then, you are free to do what you like. I seriously doubt that is how it is set up, so there's no point in going further on that topic.
So, why would one machine always stay logged in longer than another? It seems weird. Is it actually selection bias? The time to stay logged in usually depends on usage, not the machine itself. Are the machines being used differently? If so, the usage is what is causing the difference in timeout. -- kainaw 03:07, 10 August 2010 (UTC)[reply]
It would be very unlikely to do this at the DNS level, as once granted access, it's not easy to convince a browser to use the 'real' IP address due to caching and persistent connections. Typing in IP addresses will not help you either, as all but the largest websites will use Virtual hosting to put more than one website on an IP address - if you just type in the IP address, the webserver may not know which website you actually want to see.
Things like this are either implemented using a (transparent) proxy server or indeed, local software. To get around it (if possible at all) you would need to know what exactly is being used. A standard proxy server can be avoided by changing configuration (if allowed) or using a different browser that you can configure. A transparent proxy server, which is a firewall that intercepts traffic on port 80, is only avoided by using an outside proxy, although reaching that outside proxy may also be blocked by a firewall. Unilynx (talk) 05:52, 10 August 2010 (UTC)[reply]

Thank you Kainaw and Unilynx.

Rather surprisingly, most of us are allowed to select and set up our own computers as we wish (the "sensitive" stuff is all done by other people on some other LAN that doesn't concern me, and the rules there for hard/software are strict/paranoid indeed). Thus right now I'm typing this in my office on a Dell that I bought, whose Win Vista I replaced with Kubuntu, and whose (suboptimal?) settings I can't blame on anybody else.

It's fair for me to have to log in two or three times a day (or however many times I wake up a computer or turn on a new one), but what's mighty irritating is having to do this a lot more often. So I don't want (or don't think that I want) to find any way around any systemic requirement; I just want to reduce the number of irritations.

Oh, attempts to browse show that I've been kicked out. I log on again, but this time take the trouble to read the screen and see that it's about DHCP.

My little box ("fast ethernet switching hub", its fascia tells me) was getting a bit erratic. (Authentication wasn't an issue here; instead, the computer would announce that it wasn't connected to ethernet.) So yesterday I replaced it with something that's similar but also has wireless capability. (Yippee, no more cables!) With the old hub, I could surf all day. Now I have to do the authentication rigmarole every few minutes. But the problem isn't the computer in itself, as this happily works elsewhere in the institution (using this or that other wireless router intended for people like me, or piggybacking) for hours. Perhaps there's something not quite right with the wireless router. But the fact that two apparently identical computers can have their IP numbers withdrawn after different intervals when sharing the same router (not the same model of router, but the same router) has me mystified. -- Hoary (talk) 06:51, 10 August 2010 (UTC)[reply]

There are a lot of things that could be going on. There are dozens of ways to restrict access, some better than others. The fact that your login to login time appears to be random is interesting, makes me think that perhaps it's on a system-wide "reset" style timer. The one thing that will identify all computers on the network (these days) is your MAC address, although it's possible to modify that. There are hundreds of access control schemes, some rely on DNS, some rely on port blocking, some do it with ip filters (true firewalls), others do it with level 2 (mac address level, things like 802.11x), and the best do it with strong encryption. Shadowjams (talk) 09:28, 10 August 2010 (UTC)[reply]
But it's not random. Sorry, I can't be bothered to use a timer, but with the computer I'm using today I had to make a new DHCP request twice merely in order to complete my previous message here. And this has been normal since my change of hub yesterday. With the exact same (Kubuntu) computer, I'd have been able to surf uninterruptedly for hours using a different hub till very recently. (Same difference with either my little Ubuntu netbook or the old Fujitsu Win XP laptop I sometimes have to use here.) Perhaps it's a matter of the hub -- they're cheap enough, and buying a third one wouldn't be irritate me much. However, if it is the hub, I don't want to throw my money around at random; I want to know what to look for in a replacement. (The boxes advertise encryption standards and download speeds, never, as far as I can remember, good performance at hanging on to IPs granted by DHCP.) But hang on, no: attached one single hub in another office, one wintel is significantly less irritating than an apparently identical wintel; so if it's the hub, it's not only the hub. Umm..... -- Hoary (talk) 11:26, 10 August 2010 (UTC)[reply]
Surely your large institution has some sort of helpdesk where you can take issues like this? The range of possibilities here are mind-boggling. To put it simply, if you only have one connection coming into your office or area and you put a Fast Ethernet switch on it, the network at-large has no idea the switch is there (other than it sees multiple computers at the same time on the same ethernet port). If you use an off the shelf wireless router/switch and plug that cable into the WAN side, all of a sudden there is a big difference in your connection, the device is doing NAT to hide your IP address from the network at-large. I would recommend against this. Turning to your browser log-in issues: does the login screen advise you to leave the page open after you log in? Some connection systems rely on the browser window refreshing the connection to persist your authorization. If you log in and close that page you will be quickly de-authorized. Other options that can cause quirky issues include IP based authentication using DHCP, whereby your log-in moves your IP to an allowed pool for the duration of your connection; if the organization runs out of DHCP address space (such as if your two person office is connecting 5 unique machines...) it may be recycling addresses by pulling the oldest one (even if in use) and moving it to the newly connected machine, and resetting the authentication system. Really, the possibilities are endless, I could go on all day about this. Your best hope is to get ahold of your helpdesk, state your complaint and see what they have to say. Feel free to post their response if you aren't happy with it, it might give us something to go on. --144.191.148.3 (talk) 13:44, 10 August 2010 (UTC)[reply]
Hey, this is good stuff. Please do go on all day! (Ah, but then I should offer to pay you.) I haven't yet asked at the help desk because its answers to other questions tend to be carefully bland; however, I'll try asking them tomorrow. (It's now late evening here.) I think I'm doing precisely what you recommend against. No, the log-in screen doesn't recommend that you keep it open. There are never more than two machines connected in my office, and usually (e.g. today) there's no more than one connected. There ought to be no shortage of IP numbers, as most people are away on vacation. Thank you again! -- Hoary (talk) 14:45, 10 August 2010 (UTC)[reply]
It might be something as trivial and boring as a per-workstation configuration option set up in the authentication head end. If they track MAC addresses (which happens anyway when using DHCP) they can associate certain workstations with certain parameters; perhaps they want to force new/unknown stations or stations in public places to be signed out much faster, to increase security. As you described, using the firewall/NAT device caused the timeout for connected machines to drop precipitously, which corroborates this theory. If you offer your MAC address to the helpdesk, I would bet they can make a simple change to increase the timeouts for those affected machines. Good luck! --144.191.148.3 (talk) 17:08, 10 August 2010 (UTC)[reply]

Flag problem

[edit]

Hi guys. On my wiki, flagSare seeming to dissapear! Help! The first flag to dissapear is that of Côte d'Ivoire, now not only that flag is missing, but also the flag of China PR! Please reply! Velociraptor888 04:47, 10 August 2010 (UTC)[reply]

Would this perhaps be related to some recent comments on your user page? Unilynx (talk) 06:12, 10 August 2010 (UTC)[reply]
No. They just show a border with blankness. Velociraptor888 10:09, 10 August 2010 (UTC)[reply]
Hello! I SERIOUSLY need some help now. It's been 3 days since I posted. Velociraptor888 16:21, 13 August 2010 (UTC)[reply]
I've fixed it. You just need to reupload the broken file. Velociraptor888 16:24, 13 August 2010 (UTC)[reply]

Unpacking Microsoft Outlook folders

[edit]

Morning all,

My MS Outlook filing system is very simple, I put all my emails into one folder. Then I can see what emails I have received that day. As my job has become more complicated, I now have to create sub-folders for my emails, filing them per category. My question is, can I still view all recieved emails together without going through and looking at them separately? So at the end of the week, for instance, I can see all emails recieved since Monday together and sorted by date? In explorer, I can search for *.* and view all files in all the subfolders - I would like a similar solution in Outlook, is there something obvious I have overlooked?

Thanks 195.60.13.52 (talk) 09:02, 10 August 2010 (UTC)[reply]

I think I've found it, a 'search folder' would seem to fit the bill, although if you have any other suggestions... 195.60.13.52 (talk) 10:47, 10 August 2010 (UTC)[reply]

It depends which version you are using, but in my Outlook 2010 there is a "Navigation pane" on the left which shows all my mail folders in a tree structure (just like the folders in the left pane in Explorer). Any folder with unread mail is shown in bold and with the number of unread items next to the name. Of course, if you have read some of these, they won't show up as bold or be counted in the number; you will then need to use a search folder to show you just the mails based on a search criteria such as recieved date. Astronaut (talk) 09:35, 11 August 2010 (UTC)[reply]

Does anyone know how to figure out this MD5?

[edit]

If anyone knows how to decode this please tell me. This is one of my own passwords and would like to see if anyone can figure this out for me. If you can please contact me at <email address removed> Thanks 819213c9a4504af3f271ccdcc57af20b —Preceding unsigned comment added by 99.163.124.138 (talk) 10:06, 10 August 2010 (UTC)[reply]

We don't reply to external e-mail addresses, we just answer on here. Anyway, you can't "decode" MD5s, all you can do is run them through rainbow tables and hope for a brute-force match up. --Mr.98 (talk) 11:15, 10 August 2010 (UTC)[reply]
Sorry, questioner, you are mistaken when you think you can "decode" it. There are infinitely many things that have the same MD5 password, since MD5's are fixed. It's like if instead of really checking your password, the computer would hash it and get 0-9 with the MD0-9 algorithm and only accept a password that ends up with the same number. You've tried a few different passwords, but they all hash to 2, 3, 7 and 4. Now you come to us and say "Can someone help decode my password? It hashes to 9." Well, there are infinitely many passwords that hash to that number! So, you can never recover your original - you can just recover another one that will work too. 92.230.233.247 (talk) 14:13, 10 August 2010 (UTC)[reply]
You can use rainbow tables to get MD5 values. The difficulty is whether or not the original is too long or not — if it is not susceptible to a dictionary attack. You cannot decode a MD5 by working backwards, but you can find out what likely created them, if you have the time/computing resources. The computing time goes up exponentially with every additional digit, however. (All of which is made damningly near impossible if an unknown salt is added.) --Mr.98 (talk) 23:01, 10 August 2010 (UTC)[reply]
A counting argument suggests that (assuming the password's length (in bytes) is much smaller than the MD5 sum (in bytes)) there's probably only one password of a plausible length that hashes to that. However, I don't think that MD5 is weak enough to allow for an outright reversal of the hash. Paul (Stansifer) 03:25, 11 August 2010 (UTC)[reply]
pray tell what would that counting argument be? 92.230.232.58 (talk) 11:01, 11 August 2010 (UTC)[reply]
A md5 digest is 16 bytes long, so there are 256^16 different digests. Assuming the password consists of printable ASCII characters and is at most n charaters long there are (94^(n + 1) - 94)/93 different passwords. Birthday problem from there. It would be nice to have a table but I'm on a cell phone. --85.77.213.188 (talk) 12:41, 11 August 2010 (UTC)[reply]
As per our Cryptographic hash function the MD5 hash is designed to make it infeasible to deduce the source string from the hash. There are some known theoretical weaknesses in MD5 that might be exploitable. The earlier (and related) MD4 hash is known to be no longer secure against collisions and may be insecure against calculating the input string. CS Miller (talk) 22:12, 12 August 2010 (UTC)[reply]

Uptime

[edit]

Sometimes my internet goes down at night when they're doing maintenance on or, or perhaps even just turning it off to lower bandwidth from people who leave downloads going at night. I want a program that can monitor when the internet is working and when it goes down. I was suggested PingPlotter by someone, but you have to buy it and the free version is very restricted. Are there any free programs that can do this? 82.43.88.151 (talk) 10:40, 10 August 2010 (UTC)[reply]

If you just want to ping a website so you know when it's offline, here is an easy step: 1:Run cmd (by going on Start-->Run-->cmd. If you need admin privileges, , right click, run as administrator ), type ping (website) -n 10000000000000 So basically it's telling cmd to ping your website (I'll ping google of I were you) and the -n 100000.... is the amount of times to ping it. Assuming you are going to leave it long, you can tell it to ping a hundred trillion times. so this is what I'd enter: ping www.google.com -n 100000000 Sir Stupidity (talk) 10:55, 10 August 2010 (UTC)[reply]
No. I want to monitor MY internet connection over a period of time, so I know if it is going down at night and for how long. 82.43.88.151 (talk) 11:09, 10 August 2010 (UTC)[reply]
By pinging google, you can see whether you have an internet connection or not. However if you want to know what times it is down, I can offer no helpSir Stupidity (talk) 11:28, 10 August 2010 (UTC)[reply]
This batch file will keep a log of the time and the ping results, but it would have to be analysed by hand:
@ECHO OFF
:start
echo The time is: >>internetlog.txt
time /t >> internetlog.txt
ping -n 1 google.com >> internetlog.txt
goto start
--Tango (talk) 12:06, 10 August 2010 (UTC)[reply]
Some bash code (which will probably run under Cygwin):
#!/bin/bash
while true
do
    if ! ping -c 1 -w 1 google.com >/dev/null 2>&1
    then
        date +%s >>ping_fail.log
    fi
    sleep 10
done
This tries to ping google once every ten seconds, and if it doesn't receive a reply within a second, it writes the Unix timestamp of the failure to the file "ping_fail.log". That should be sufficiently close to what you want; simply reviewing the log allows you to see (with approximately a ten-second granularity) when the internet went down (the first time stamp in the file, provided it was empty when the script started). Moreover, if there are significant gaps between time stamps, it means the internet was back up at that point. If your internet is up and running in the morning, it means it came back up within about ten seconds after the last timestamp in the log. (N.B. this code will, of course, add about 360 time stamps to the log per hour of downtime, so clean it regularly, or it could get rather big. It should, however, be a bit nicer on the system than the batch code above. ;) ) --Link (tcm) 12:18, 10 August 2010 (UTC)[reply]
I'm not the OP, but I have a similar-ish problem (My internet disconnects every minute, for 2-3 seconds). These solutions sound ideal to offer as proof for this (my internet provider doesn't believe me). But I've got no idea what they are. I'd have to download (and learn how to use) Cygwin? What is that? And would that be difficult? Vimescarrot (talk) 14:42, 10 August 2010 (UTC)[reply]
Ah - never mind - a friend informed me what a batch file was and how to make it, and I got it running. And I was right, after a minute or so I get a single entry with a ping of 1968. Vimescarrot (talk) 15:14, 10 August 2010 (UTC)[reply]
Really sorry to be a pain but would it be possible to add a little something to Tango's code to make it add some line of text that I could easily ctrl + F if the ping is over 1000? Vimescarrot (talk) 15:25, 10 August 2010 (UTC)[reply]
Alternatively, changing the code so that it only records the details when the ping is over 1000 may shorten the file and make it easier to collate. I, unfortunately, cannot write the code, so this is just a suggestion. --KägeTorä - (影虎) (TALK) 15:29, 10 August 2010 (UTC)[reply]
I don't remember the output format of the (Windows) ping command immediately, but it should be completely trivial to search for such lines with a regular expression. --Tardis (talk) 17:55, 11 August 2010 (UTC)[reply]
Nobody has suggested checking the system logs? Windows will log an event when the network card goes down. Here are specific instructions for debugging network problems using Network Diagnostics Event Logs. Nimur (talk) 21:22, 10 August 2010 (UTC)[reply]

Netgear DG834G V3 wireless router

[edit]

Hi everyone I have the above router for my wireless broadband connection. My ISP is AOL. I had another netgear router of the same type my previous home but used a different ISP. In my previous house I had wireless internet but didn't have to pay for a BT landline. Since moving house, I tried to unplug my router from the phone line and lost my internet connection. I was wondering if there is a way that I can keep internet access if I were to cancel my BT landline, as I was able to have wireless internet using the same type of router without paying for a BT landline previously. Hope this makes sense. Any suggestions would be greatfully received. —Preceding unsigned comment added by RichYPE (talkcontribs) 11:32, 10 August 2010 (UTC)[reply]

It's only wireless between the router and your computer(s). You still need a wire between the router and the internet. When you weren't paying for a BT landline you were probably pay some other company for your landline (possibly included with something else, Sky TV maybe?). There are alternatives to using a landline for the internet (such as cable (the same cable as for TV, which may or may not be available in your area) or satellite), but they are usually more expensive than just having a landline (cable gives a faster connection and satellite can give broadband speeds when they aren't available on landlines in your area (which is very rare these days), but you pay more for those benefits). --Tango (talk) 12:13, 10 August 2010 (UTC)[reply]
You probably had TalkTalk or a similar company that offers services through Local-loop unbundling. These companies can take over all aspects of internet and phone so you no longer have anything to do with BT (apart from BT still maintaining the physical cable to your door). You can also get (expensive) internet through the 3G mobile telephone network. Astronaut (talk) 09:09, 11 August 2010 (UTC)[reply]
... but not nearly as expensive as satellite phone and internet. "Sky" might be a satellite company, but they don't provide satellite internet - a phone line is still needed. Dbfirs 20:22, 11 August 2010 (UTC)[reply]

Put similar named files into sub-folders

[edit]

Is there any freeware available that can put all files with similar file names (such as those with the first n characters the same) into the same subolder? 92.15.3.61 (talk) 15:59, 10 August 2010 (UTC)[reply]

I don't know if this fits the bill for what you are trying to do, but I can think of two ways to accomplish this in Windows.
1. Let's say you want to move all files starting with "web" to the directory "webstuff". From the command line, first cd to the proper directory and then type move web* webstuff and the files will be moved immediately.
2. In Windows, right-click the folder containing the files you are looking for. Search for web*. The window will slowly populate. When the search is complete, hit ctrl-A to Select All, then drag them all into the destination folder. Unlike technique #1, this will work for all files starting with "web" that are in the folder you originally right-clicked plus its sub-folders.
Comet Tuttle (talk) 17:44, 10 August 2010 (UTC)[reply]
Difficult to do when you have hundreds of different file groups. 92.28.253.92 (talk) 23:25, 10 August 2010 (UTC)[reply]
This is fairly easy to do on any platform, but it's impossible to answer the question without knowing what platform you're on. Mac, windows, linux, unix? software version? --Ludwigs2 21:05, 10 August 2010 (UTC)[reply]

WinXp. 92.28.253.92 (talk) 23:24, 10 August 2010 (UTC)[reply]