Topic: Problem connecting to Internet using DHCP client on eth1

[Robert]

I've seen several postings regarding this problem and I've tried the recommendation offered in the bug reports under the heading "ISP DHCP response corrupts resolver configuration".  I simply cannot get the server to sync up with my cable modem (COMCAST).  I was using 2 3com cards but, then I switched to using just one and an HP2585B card under the recommendation from a tech support guy at Comcast who had experience w/ Linux.  I've tried several different PCI NIC cards and this is just not working and now I'm looking for any help I can get.  It seems to me that this is pretty fundamental and I shouldn't be having the kind of trouble I'm having.

When I run ifup eth1 it gives me the following message:
"Determining IP information for eth1...  failed"

I'm sure others have had this problem so, if you were able to get past it, PLEASE pass on your words of wisdom.  A few details:
The hardware platform is a 133MHz HP Vectra with a 2GB disk and 2 NIC cards.  The 3com 3C905B is on interrupt 9 and the HP card is on interrupt 11.  The HP card is being used as eth1 to connect to the external network.  I've tried connecting to the external network by passing it a client ID (didn't request a password) and by sending it the address.  Neither way works!

Any help you can offer would be much appreciated!

Thanks,

Robert

[Dub Dublin]

I have a VERY similar problem that I've been wrestling with all night - here are the details:

I have been using SME 5.1.2 as a server only for several weeks for pre-deployment testing.  Ultimately, it will need to operate in gateway mode, too, so when my Linksys cable NAT/router box bit the dust this morning, I decided this would be a good time to slap another NIC int he box and get some stick time on the gateway.

No matter WHAT I do, the SME server will NOT get a DHCP address from Roadrunner (to be honest, I'm not even sure if that addr is handed out by the router back at the headend, or by my Toshiba cable modem locally.  It works fine though, because I just plugged the cable directly into my PC to post this.

I may also be confusing it slightly with the DNS setup (see below), but I really don't think that should be stopping things, since the failure to acquire an IP address, etc. through DCP clearly takes place well before DNS is ever awakened.  Following are all the gory details.  ANY AND ALL HELP OR PONTERS APPRECIATED.  Thanks.

Here's the SME Config:
********
Primary Domain Name: "austin.rr.com"  (Not sure if this will choke things, but I want it to use external DNS, so I think this is sort of right.  If not, again pointers are appreciated.)

System Name: "cyberia" (not like it should matter...)

Local Ethernet Driver: "Bridgecom, (driver tulip)" (A Linksys 10/100 card I added today.  Validated as working hardware when internal and external interfaces are swapped.)

Local IP Addr: "192.168.1.1" (Also tried the default 192.168.248.1, no difference.)

Local Subnet Mask: "255.255.255.0"

Operation Mode: "Private Server and Gateway"  (Seems to make no difference whether I pick the Private variety or not, the failure mode is the same...)

External Access Mode: "server and gateway - dedicated"

External Ethernet Driver: "Intel Corporation|82557 [Ethernet Pro 100] (driver eepro100)" (Again, hardware validated as working - it makes no difference which interface is which, if I switch them around, the external one will always fail to get its DHCP address, but the internal one comes up fine.)

External Interface Configuration: "DHCP with Ethernet address" (Although I'm sure this is the right choice, I've tried the account name option, both blank and with a bogus account - still doesn't work. I'm on Roadrunner with a Toshiba cable modem, if that helps - no PPPoE.  The Linksys box and my Win98 box grab addrs just fine with their default configs.)

Dynamic DNS Service: "off"

DHCP Server Configuration" "on"
Beginning DHCP Host Range: "192.168.1.65"
Ending DHCP Host Range: "192.168.1.250"

Master DNS Server Address: "24.93.35.32"  (This is the current setting, as it's one of the DNS servers for austin.rr.com.  I've also tried this setting as blank, no difference.)

Proxy Server: "No"

Status Reporting: "No" (This better NOT be the problem!)

Console Mode: "auto"

Contact e-mail address: "" (undefined)
********

The failure mode is as follows:  The machine comes up, and hangs for a while on the Starting Network item, before continuing.  Whichever interface is defined as eth1 will fail to ever have its IP address set, as shown by ifconfig -a:

eth0      Link encap:Ethernet  HWaddr 00:04:5A:74:1C:C3  
          inet addr:192.168.1.1  Bcast:192.168.1.255  Mask:255.255.255.0
          EtherTalk Phase 2 addr:65280/161
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:4 errors:1 dropped:0 overruns:0 frame:2
          TX packets:93 errors:12 dropped:0 overruns:0 carrier:24
          collisions:0 txqueuelen:100
          Interrupt:11 Base address:0xe400

eth1      Link encap:Ethernet  HWaddr 00:40:CA:11:27:BD  
          BROADCAST  MTU:1500  Metric:1
          RX packets:124 errors:0 dropped:0 overruns:0 frame:0
          TX packets:4 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:100
          Interrupt:10 Base address:0xe800

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          EtherTalk Phase 2 addr:0/0
          UP LOOPBACK RUNNING  MTU:3924  Metric:1
          RX packets:6 errors:0 dropped:0 overruns:0 frame:0
          TX packets:6 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0


Likewise (at the risk of providing too much meaningless related data), /var/log/messages shows the following, which pretty much fits, since it's obviously NOT getting the external IP address via DHCP:

Mar 30 17:44:35 cyberia network: Setting network parameters:  succeeded
Mar 30 17:44:35 cyberia network: ^[[60G
Mar 30 17:44:35 cyberia network:
Mar 30 17:44:35 cyberia network: Bringing up interface lo:  
Mar 30 17:44:36 cyberia network: Bringing up interface lo:  succeeded
Mar 30 17:44:36 cyberia network: ^[[60G
Mar 30 17:44:36 cyberia network:
Mar 30 17:44:36 cyberia network: Bringing up interface eth0:  
Mar 30 17:44:36 cyberia kernel: tulip.c:v0.91g-ppc 7/16/99 becker@cesdis.gsfc.nasa.gov
Mar 30 17:44:36 cyberia kernel: eth0: ADMtek Centaur-P rev 17 at 0xe400, 00:04:5A:74:1C:C3, IRQ 11.
Mar 30 17:44:36 cyberia network: Bringing up interface eth0:  succeeded
Mar 30 17:44:36 cyberia network: ^[[60G
Mar 30 17:44:36 cyberia network:
Mar 30 17:44:36 cyberia network: Bringing up interface eth1:  
Mar 30 17:44:37 cyberia kernel: eepro100.c:v1.09j-t 9/29/99 Donald Becker http://cesdis.gsfc.nasa.gov/linux/drivers/eepro100.html
Mar 30 17:44:37 cyberia kernel: eepro100.c: $Revision: 1.20.2.10 $ 2000/05/31 Modified by Andrey V. Savochkin and others
Mar 30 17:44:37 cyberia kernel: eepro100.c: VA Linux custom, Dragan Stancevic 2000/11/15
Mar 30 17:44:37 cyberia kernel: eth1: OEM i82557/i82558 10/100 Ethernet, 00:40:CA:11:27:BD, I/O at 0xe800, IRQ 10.
Mar 30 17:44:37 cyberia kernel:   Receiver lock-up bug exists -- enabling work-around.
Mar 30 17:44:37 cyberia kernel:   Board assembly 664088-003, Physical connectors present: RJ45
Mar 30 17:44:37 cyberia kernel:   Primary interface chip i82555 PHY #1.
Mar 30 17:44:37 cyberia kernel:   General self-test: passed.
Mar 30 17:44:37 cyberia kernel:   Serial sub-system self-test: passed.
Mar 30 17:44:37 cyberia kernel:   Internal registers self-test: passed.
Mar 30 17:44:37 cyberia kernel:   ROM checksum self-test: passed (0x24c9f043).
Mar 30 17:44:37 cyberia kernel:   Receiver lock-up workaround activated.
Mar 30 17:44:37 cyberia ifup: Determining IP information for eth1...
Mar 30 17:44:37 cyberia dhcpcd[684]: broadcasting DHCP_DISCOVER
Mar 30 17:45:37 cyberia dhcpcd[684]: timed out waiting for a valid DHCP server response
Mar 30 17:45:37 cyberia kernel: eth1: 0 multicast blocks dropped.
Mar 30 17:45:37 cyberia ifup:  failed.
Mar 30 17:45:37 cyberia network: Determining IP information for eth1... failed.
Mar 30 17:45:37 cyberia network: Bringing up interface eth1:  failed
Mar 30 17:45:37 cyberia network:
Mar 30 17:45:37 cyberia rc: Starting network:  succeeded

Again, any and all help would be appreciated.   I've been doing Internet networking since 1985, so I've got a clue what I'm doing , but this appears to be a RedHat/SME problem to which the answer is not real obvious...

[Des Dougan]

Dub,

I use an ADSL service, rather than cable, but the principle may be the same... My ISP requires that the MAC address be registered before it will hand out an IP address. Is it possible that you've previously registered the PC's address with them in the past? Maybe you could try the PC's NIC for the external interface?

Also, you should only set up as Private Server and Gateway if you want the following:

Your web server is not visible to anyone outside of the local network.
Your mail server is not accessible from outside of the local network.

HTH,

Des Dougan

[Dub Dublin]

DOH!  Actually, RoadRunner doesn't require registration of MAC addrs, but I figured out what tripped me up.  If it looks like there's nothing wrong in the config up above, that's because there isn't anything wrong.

The problem is that a couple of months ago, RR replaced my older Motorola cable modem with a Toshiba.  It turns out that although the MOT unit supports up to 5 dynamic IPs locally, the Toshiba only allows ONE.  This was never apparent since I was running a Linksys NAT/Router box, and so they only saw a single dynamic IP addr.  (I've since been told I should have held on to the MOT modem.  Too late for me, but maybe not for you.)  DNS (!) and everything else seems to be working as expected.

The fix was simply to turn everything off, then turn on the cable modem, the SME server, and the client PCs (in that order, of course, nothing else would make sense.)  The cable modem then forgot all about the IP it had handed out before, and life is good.  At the mercy of stupid Japanese hardware, but good...

[Robert]

You know, I've tried turning the cable modem off for 5 minutes and then turning it back on again.  I've called COMCAST to have them provision the cable modem.  They had the wrong MAC address in their database.  I've tried using several different NIC cards but, nothing, thus far, seems to work.  I'm just about ready to give up and use plain vanilla Red Hat 7.2 or get a Linksys router/modem.  I hadn't posted here before so, I wanted to give this avenue a chance.  The SME server came highly recommended so, I wanted to give it a try, but, it seems that me that in the performance of it's principle function of providing connectivity between one or more networks, it's failing miserably.  I know that others have had this problem so, I'm wondering why a patch has not been provided.  If there is one, please provide a link.

Thanks,

Robert

[Liam]

this is not a peronet solution but it should work untill u get it figured out. when ur server doot up to the counsole hit alt+f2 and login under root with the admin password.
when the promt comes up type    dhcpcd and hit enter.

go back to the counsole by hiting alt+f1and test ur internet.

it should be working

[Robert]

Sorry to say that didn't work either!  I'm afraid I'll just have to tinker with it at a later date because I've been trying to get this to work for the last couple of weeks and I have had no success.  I was encouraged to use the server because I was told it's very straightforward and effective.  I've found the install process itself to be very straightforward and simple, but it is not working for me so, I've got to go with some other solution.

Thanks for your suggestions!

Robert

[john]

Hi Robert,
Looking for a solution to my problem, I came across your thread - I actually might know the solution - when you configure e-smith for Comcast DHCP, on the screen labeled "External Interface Configuration" you need to choose item # 1 - "Use DHCP [send account name as client identifier" so that you will be authenticated by them. The client identifier is your Comcast hostname. I've connected a number of machines successfully, with that being the only real requirement. good luck!

[Robert]

Hi John,

Thanks for your reply.  When you say hostname, do you mean the username?  If so, I've tried that to no avail.  In general, I can't seem to connect at all although both NIC cards have been confirmed to be functional and they are on different IRQs.  Eth1 won't come up for some reason.

Thanks again!

Robert

[john]

Hi again,
No the hostname is usually different from the username. The hostname is usually something like "cn-377271-a", at least in my area. If you have been using the correct name & syntax, the problem might be with the local Comcast database - I had a lot of problems with my personal account because of a flaky local server/database. You might also have an obscure hardware problem - would it be safe to assume you have tried reversing the nic cards in the configuration, so that eth0 becomes the external port & eth1 the local port? That will confirm that everything about the nic assigned as eth1 is all right!
To quickly resolve a nagging problem, I have had to occasionally replace both nic cards or the pc itself because of some obscure hardware problem; I think the SME server software is great - but like all software an obscure bug will sometimes surface - the best way to avoid those bugs or fix them is to be prepared to quickly change hardware or simply re-install. I would suggest the following, in sequence:
1] replace both nic cards [3Com is my favorite - they always seem to work]
2] replace the pc with a known good system of recent manufacture, with *new* nic cards.
3] If the problem still persists after those actions, it is definitely a Comcast problem. You can get them to address it directly by connecting up a win98 machine directly to the cable - if it can't connect - their tech support will be forced to schedule a site visit to diagnose the problem [I once had a problem with flaky cable voltages-that was the only way it was found].

Sorry I can't suggest a better resolution. Good luck!

[mike mattos]

just a dumb newbie idea, are you sure it is dhcp and not PPPoE ?  Around here ( toronto ) pppoe is very common.

also, with some cable providers, their SERVER maps your MAC address, if you don't log out, the server will not allow a reconnect from a different MAC address, bit awkward when your network adapter dies!  You have to phone support and they release the address lock

OTOH, if you politley release from the router/gateway, the new equipment logs in fine

ALSO,. the NAME of the login host has to match their assigned name, again a bit awkward!  

2 cents worth!

mike

[mike mattos]

just a dumb newbie idea, are you sure it is dhcp and not PPPoE ?  Around here ( toronto ) pppoe is very common.

also, with some cable providers, their SERVER maps your MAC address, if you don't log out, the server will not allow a reconnect from a different MAC address, bit awkward when your network adapter dies!  You have to phone support and they release the address lock

OTOH, if you politley release from the router/gateway, the new equipment logs in fine

ALSO,. the NAME of the login host has to match their assigned name, again a bit awkward!  

2 cents worth!

mike