Linux - NetworkingThis forum is for any issue related to networks or networking.
Routing, network cards, OSI, etc. Anything is fair game.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
My (oh so kind) uncle was kind enough to donate two Realtek ethernet cards to me.
This was because he'd promised me just one, but he thought one of them was dead.
I'm trying to connect two boxes to each other, got TCP/IP working, they can ping themselves, but not each other. Beacause the Linux box says 'destination host unreachable' and the win95 box says 'request timed out.'
I tried plugging the crosslink cable I'm using into a (tested) working eth card, and that didn't work too. The link/act is lit every time I plug in the cable on both boxes, so the cable shouldn't be the cause.
I've put both eth cards into my linux box, which it could detect. Win98 had no trouble at all detecting it, too ...
So could there be a 'dead' eth card? Or is an eth card never 'dead' when your OS can see it?
It is possible that it is dead. Usually a good indication of a working card is to ping it's own ip address from itself. ie on a box with a card use 192.168.1.1:
ping 192.168.1.1
If you geta response then the card is "probably" good.
It usually means that it doesn't know where to locate the host on the network. This can usually be overcome by adding a route to your routing table:
route -n
Hm. I haven't seen 'route' before, /me go check man route.
Edit: I've once connected two linux boxes with (two) straight though cables by means of a hub. That worked perfectly, simply define the IP address, plug in and go play Quake.
I thought this would work out the same way, but it wouldn't. How come I do need a routing table for this network, and perhaps not for my previous one?
Distribution: Slack Puppy Debian DSL--at the moment.
Posts: 341
Rep:
I just had a realtek based card do inexplicable things to my network--untrace-able weird and annoying things--like I thought my router was going south because I had to start swapping ports. When the card finally bit the dust and was replaced all of the intermittent problems disappeared.
It was probably due to inproper handling, I really mistreated that card, although I also live in Tampa Bay area--one of the highest lightning activity areas in the world. Who knows?
The point I am trying to make is: a NIC doesn't have to be "dead" to be "BAD". If the bad one wasn't labeled, spend the $20 and get a "known-good" NIC and pitch them both. It isn't worth the trouble for 20 bucks.
Also, when connecting directly, computer-to-computer ( when not using Win2000 or XP to same,): Go to the windows box, find HOSTS.SAM copy it to HOSTS in the same directory and make an entry for the linux computer, then make an entry for the windows box in the linux Hosts file. [ ip-address computer-name] Assign static Ip-Addresses for both machines for their NICs connected to one another. I don't think you have to set up a private domain, just use private network addresses {ie. 192.168.123.101}.
The "hub" that you were talking about must have been a router, not a network switch. A router acts as host for a local domain, and performs some other services as well. A switch makes multiple connections possible to a network, but provides no services.
[[ I recently had a Network Interface Card with damage from inproper handling (static discharge damage) which pinged itself and other locations on the network--yet it corrupted data; had weird (and untrace-able), intermittent problems; and somehow affected the whole network (The router problem I talked about earlier.).
Sometimes, Electro-Static Damaged PCI adapter cards cause all sorts of mayhem internally as well: Stack corruption; Buffer under/over-runs; Weird intermittent memory corruption; All sorts of stuff which may or may not lead to fatal errors (Blue Screens, and the "This program has performed an illegal action and will be shut down--Please Pardon the inconvenience".)
That is why all of the manuals are so hysterical about proper handling of any adapter card for a computer. Many cards are too tough to die outright--They may never die or will linger on painfully (leading to the need for mental health counseling for anyone who has to use the computer or the network.)
I personally made a total ass-clown of myself over "Software errors". As in: spending hours with Microsoft, bitching about the last patch hosing my network--only to have an adapter card die and all of the problems disappear.
Any device which has it's own internal buffers and DMA--what I call an "Intelligent" or "Semi-intelligent" adapter card--can hose a system if it is damaged. It can be subtle errors which are darn near impossible to trace. It can be glaring problems which appear to be coming from someplace else. That is why everyone tells you to start with hardware first, when trying to track down a problem.
I've also hosed things up by not reading all of the directions when updating bios on certain devices--like the time I updated the firmware on a DVD without unplugging the hard drive on the same IDE channel ( I don't trust the hard drive anymore, although it appears to be okay, but the DVD has been "goofy" ever since--even after reflashing it.).
Sometimes, if you are lucky, updating the firmware on a problematic device will be a "kill-or-cure". Read all of the directions; unplug everything you can or do it on a stripped "lab machine"; and double-check to make sure you have the right bios for the device_before_you_flash_it.
Label the card or device as one you had to update and put a label on the inside of the case with everything you did--in case: A) The logs get mis-placed. B) You didn't log it in the first place. (Would I do such a thing? N-a-a-a, I ALWAYS remember to log everything. Sure I do.)
Why did I write such a long and rambling answer? Because I need to remind and reinforce myself on stuff I already know. I have gotten away with sloppy habits too many times--the nightmare NIC episode lasted for months and most likely could have been avoided. At the very least, I could have been sure it wasn't ME--I have been kicking myself for weeks over it.]]
I'm gonna try your conf right now, thank you.
I was wondering: even if my boxes are unplugged, they give the precise results I described earlier. How come?
Distribution: Slack Puppy Debian DSL--at the moment.
Posts: 341
Rep:
Every network/domain has to have a host. Just as your computer has to define itself as "local-host" at 120.0.0.1 on "local-domain" at 120.0.0.0 for its own inter-process communication. This is true of any network-capable O.S..
A router is the host for an external domain, like a LAN:
It sends requests on the LAN for all connected devices to identify themselves (ARP requests);
Verifies MAC addresses;
Provides IP-addresses to devices via DHCP;
Advertises available services (like print-servers);
It announces the available protocols running on the network (so that all of the devices use a common language to announce the available services provided by each member of the domain--like shared folders or drives and other connections);
It sends out the commands which enable most of the network settings to be dynamically assigned by each O.S..
Tcp/Ip is a packet-protocol which encapsulates other protocols (like http, SMB, and others). The host of the local network also sends packets in other protocols in addition to the Tcp/Ip; an example is the SNMP--Simple Network Management Protocol.
Windows 2000 and XP have many of these services running in the background. They do auto-negotion to determine which unit is going to provide which services on a network, generate ARP requests, and announce available shares and additional services. So when you connect either of those O.S. together, it is done "auto-magically". Mostly. They won't do it for Linux/UNIX or "older" non-supported Microsoft products.
Whoa! First off the loopback IP is 127.0.01! Not some other number! This would be the first number to ping to see if your card is good. If that doesn't work you will need to trouble shoot drivers, card hardware and system config(on a Linux box). If you can't get the loopback IP to work on your Linux box I would put it in a Windows box first before tossing it. You could have your Linux box configured incorrectly. I'm not sure of the files to edit at this second. If you can ping the loopback now you are ready to forge ahead. You need a CROSSOVER cable to connect two NICS directly. If you are not absolutely sure it's a crossover get one that is sold as such. LABEL it! Don't mix it up with your standard cables. Next set up a simple internal IP scheme of 10.0.0.2 with a subnet of 255.255.255.0 and the other with the IP of 10.0.0.3 and the SAME subnet. Once again it's not fresh in my mind which files to edit. Once you do this you should be able to ping. If you can't wait say 30 minutes with both machines running to let the ARP tables update in case you had different IP to MAC mappings in there then try again. If you do buy new cards get 3Com network card. 3COM ETHERNET Adapter Model 905C-TXM PCI 10/100M would be good or 905B's are good to. Try somewhere like newegg.com.
Distribution: Slack Puppy Debian DSL--at the moment.
Posts: 341
Rep:
What whoa! I don't think he changed that number. He may be reporting the first address. He doesn't need to set up a whole domain and frickking DNS to connect two computers with X-over cable. 10.0.0.0 is normally used for internal addressing on larger networks like in your computer lab. 192.168.123.2-254 is used for small private networks. That's standard.
If you read the thread from the start, you would know he just wants to connect with a X-Over cable which he already has. He has two cards given by an uncle--one is supposedly "bad" they are unlabelled and loose. Neither is outright "dead". For non-production environments there is no reason to spend more than twenty bucks at a local store for a NIC.
[Back to the thread]
OK. The windows Host.sam copied & renamed to Hosts should have those entries.
However, the host's names (ie. "winbox") isn't arbitrary, it is the name of the computer--just like in windows. When you browse the lan for members of a "Workgroup" you are looking for the computer's reported name--something you generally assign when doing the install. This is true in Linux also. You aren't setting up a work group or realistically a domain. If you use the correct hostname for the windows box, you should be able to ping it through the cross-over cable.
If you have 98se or ME (Yuck) or W2k, or XP on another computer.
think about the "network wizard". The questions it asks and the process it follows is the same as you are doing manually in the hosts files in the respective machines. Windows will put the information in the registry as default using the wizard--but it will read a hosts file before looking anywhere for DNS information.
Your machine has the loop-back--which you already knew because you didn't change anything that was already present in the file. It has a name which is assigned during install (for windows always, linux depends on the distro).
I'm cutting you off, I'm out of cigarettes and need sleep. Go to redHat online and read the Installation Guide and the Customization guide. I will give you a hint: man hostname.
Another hint: man hosts. Check and double check syntax and typos. It's late; go to bed.
[[Sorry, I really need to go to bed--I was taking the naming scheme as the actual names instead of generic. I appologize, for I am crispy brown toast--bye]]]
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.