Messed up my "network"

[nodough]

Re: Messed up my "network"

Sigh! Well, I bit the bullet and decided to try sudo chattr +i /etc/resolv.conf

And this is what I got:

sudo chattr +i /etc/resolv.conf
chattr: Inappropriate ioctl for device while reading flags on /etc/resolv.conf

And, of course, my resolv.conf was overwritten once again.

Googling tells me the error message happens often, and that there's a bug in Hardy https://bugs.launchpad.net/ubuntu/+s...rm/+bug/257840 that for all I know could be part of the problem, although the reporter of the bug is having hard drive problems.

UPDATE:
FWIW, I tried changing resolv.conf to add the correct IP addresses, then tried connecting the boxes again. They didn't connect, although the resolv.conf still had the right addresses. Now that I'm online, it's back to the 127.0..0.1

[GreyGeek]

Re: Messed up my "network"

.....
UPDATE:
FWIW, I tried changing resolv.conf to add the correct IP addresses, then tried connecting the boxes again. They didn't connect, although the resolv.conf still had the right addresses. Now that I'm online, it's back to the 127.0..0.1

I'm confused.

How did you get "online" and with what box? The one with the faulty resolv.conf file? You changed resolv.conf to hold the dns addresses of your ISP, you got online, but after you get online and then look at the contents of resolv.conf it contains only "127.0.0.1"? Yet, somehow, somewhere, a domain name is being converted to an IP address, enabling you to browse the internet normally, without timing out? Something, somewhere else, is doing the domainname --> IP address conversion for your browser. Resolv.conf is not.

You do know that 127.0.0.1 is also called the "loopback" address, abbreviated "lo". When I issue "ifconfig" in a Konsole here is what I get, with "lo" highlighted in bold:

jerry@vgnfw140e:~$ ifconfig
eth0 Link encap:Ethernet HWaddr xxxxxxx
UP BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:0 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)
Interrupt:16

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:146460 errors:0 dropped:0 overruns:0 frame:0
TX packets:146460 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:9548989 (9.5 MB) TX bytes:9548989 (9.5 MB)

wlan0 Link encap:Ethernet HWaddr 00xxxxx
inet addr:192.168.1.101 Bcast:192.168.1.255 Mask:255.255.255.0
inet6 addr: fe80::216:eaff:fe4d:58aa/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:198162 errors:0 dropped:0 overruns:0 frame:0
TX packets:121402 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:191755023 (191.7 MB) TX bytes:22483176 (22.4 MB)

I've blocked out my HWaddr's with "xx" for security reasons. What does "ifconfig" show on your box (and don't show your HWaddrs)?

"lo" is the "default" address all IP packets are sent to when no other destination can be found. If you type "localhost", or 127.0.0.1, into the URL of FireFox you'll get a "Problem loading page" error. IF you install Apache onto your system it will create a /var/www/... directory structure which will lead localhost to display the Apache message page. (And if you make it world readable and publish your IP address you have your own online presence, IF you replace Apache's index.html page with one of your own desigh.

[nodough]

Re: Messed up my "network"

.....
UPDATE:
FWIW, I tried changing resolv.conf to add the correct IP addresses, then tried connecting the boxes again. They didn't connect, although the resolv.conf still had the right addresses. Now that I'm online, it's back to the 127.0..0.1

I'm confused.

I don't blame you

How did you get "online" and with what box? The one with the faulty resolv.conf file?

Sorry I wasn't clear. My brain is quite addled at the moment, from trying to resolve my resolv.conf problem. (Then again, it was probably addled when I began this internet sharing project, hence the mess-up.)

Hardy box:

• working dialup serial modem
• ethernet card
• corrupted resolv.conf

Jaunty box:

• no modem
• ethernet card
• everything working, AFAIK

You changed resolv.conf to hold the dns addresses of your ISP, you got online ...

No, sorry. The corrupted resolv.conf seems to make no difference to getting online with the serial modem. Which makes no sense to me, but I've learned I understand even less of this than I thought.

Here's what I did, step-by-step--all on the Hardy box:
1. Not connected to internet. i.e., hit Disconnect in KPPP.
2. Change resolv.conf to correct DNS
3. sudo ifconfig eth0 up (it had been down)
4. Try to connect to Jaunty box via FTP--didn't work.
5. Looked at resolv.conf to see if anything changed there. It didn't.
6. sudo ifconfig eth0 down (to be able to use dialup modem)
7. get online.
8. Check resolv.conf. It changed back to 127.0.0.1

Which was probably a useless exercise, but made things clearer in my head.

You do know that 127.0.0.1 is also called the "loopback" address ...

Yes. It's one of the few things I understand about this. It's also the DNS of localhost, isn't it?

What does "ifconfig" show on your box (and don't show your HWaddrs)?

From what I can tell, what's below is all as it should be--what it was before I began messing up. Keep in mind that *before* I get online with my dialup modem I have to take eth0 down, so it doesn't show in ifconfig. But I put it up to show you.

eth0   Link encap:Ethernet HWaddr XX:xx:xx:xx:xx:xx
     inet addr:192.168.0.130 Bcast:192.168.0.255 Mask:255.255.255.0
     UP BROADCAST MULTICAST MTU:1500 Metric:1
     RX packets:0 errors:0 dropped:0 overruns:0 frame:0
     TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
     collisions:0 txqueuelen:1000
     RX bytes:0 (0.0 B) TX bytes:0 (0.0 B)
     Interrupt:17 Base address:0xc800

lo    Link encap:Local Loopback
     inet addr:127.0.0.1 Mask:255.0.0.0
     UP LOOPBACK RUNNING MTU:16436 Metric:1
     RX packets:161844 errors:0 dropped:0 overruns:0 frame:0
     TX packets:161844 errors:0 dropped:0 overruns:0 carrier:0
     collisions:0 txqueuelen:0
     RX bytes:85532061 (81.5 MB) TX bytes:85532061 (81.5 MB)

And, with eth0 down a few numbers change, but not the significant ones:

lo    Link encap:Local Loopback
     inet addr:127.0.0.1 Mask:255.0.0.0
     UP LOOPBACK RUNNING MTU:16436 Metric:1
     RX packets:162094 errors:0 dropped:0 overruns:0 frame:0
     TX packets:162094 errors:0 dropped:0 overruns:0 carrier:0
     collisions:0 txqueuelen:0
     RX bytes:85720800 (81.7 MB) TX bytes:85720800 (81.7 MB)

BTW, this is all done when I'm offline. I try to answer the posts offline because it's less harrowing, knowing I'm not tying up my phone. Then I paste my responses into the reply box.

When I'm online, I get ppp0 in ifconfig (I didn't block the ISP's DNS numbers, because they're posted here anyway):

ppp0 Link encap:Point-to-Point Protocol
     inet addr:134.117.137.81 P-t-P:134.117.137.5 Mask:255.255.255.255
     UP POINTOPOINT RUNNING NOARP MULTICAST MTU:1500 Metric:1
     RX packets:215 errors:0 dropped:0 overruns:0 frame:0
     TX packets:224 errors:0 dropped:0 overruns:0 carrier:0
     collisions:0 txqueuelen:3
     RX bytes:151566 (148.0 KB) TX bytes:24990 (24.4 KB)

IF you replace Apache's index.html page with one of your own desigh.

I've never tried to do Apache--and, if my current messup is any indicator, I doubt I ever will.

[GreyGeek]

Re: Messed up my "network"

OK.

Your Hardy box is connected to a phone company (your ISP) via serial modem connected to your COM1: serial port. Right?

And you are running "sudo ifconfig eth0 up" against your ethernet port? It's been many years since I last used a serial modem but when I did I used Kppp to make the connection, not the "ifconfig" command you used. Do you have your serial modem plugged into your eth0 port? If so, that shouldn't be. You need to set either PAP or CHAP authentication, among other things.

HERE is a manual for KPPP, and here is the most recent FAQ to peruse for late changing facts.

[nodough]

Re: Messed up my "network"

GreyGeek, I've probably bogged you down with way more information than you need, and confused things in the process. For that I apologize.

In a nutshell:

1. My serial modem works fine, and connects to my ISP just fine. It's plugged into the serial port, as it should be.

2. The only time I would put up eth0 is to connect my 2 boxes via the crossover cable, so that I can transfer files from one to the other.

3. The boxes can ping each other.

4. I gave you all those ifconfig variants because I have no idea what you need to diagnose why my Hardy box (using proftpd, which is running) is refusing connections from the Jaunty box.

5. We've identified a specific clue: the etc/resolv.conf file is being overwritten.

6. Running sudo chattr +i /etc/resolv.conf returns:

chattr: Inappropriate ioctl for device while reading flags on /etc/resolv.conf

and I have no idea what that means or why it's happening. Yes, I googled it, but what I found all has to do with resolv.conf resetting on boot, and people who use routers. Mine resets as soon as I go online with the serial modem, and I have no router.

7. It's possible that my problems are being caused by the configuring I did after installing dnsmasq, dnsmasq-base and resolvconf. I'd like to try uninstalling those programs, to see if things return to the way they were. But I'm terrified I'll screw up my internet connection if I uninstall them.

I hope this clarifies things for you.

[oshunluvr]

Re: Messed up my "network"

seems to me that if you're only connecting the two boxes and they ping each other AND resolv.conf isn't required by your modem, then why are we messing with it?

If you can ping each box from the other you should be able to enable services like NFS, Samba, ssh and ftp.

[nodough]

Re: Messed up my "network"

seems to me that if you're only connecting the two boxes and they ping each other AND resolv.conf isn't required by your modem, then why are we messing with it?

Because the two boxes won't connect, and resolv.conf is the only thing I've discovered so far that I know for a fact ain't how it was before the screw-up.

If you can ping each box from the other you should be able to enable services like NFS, Samba, ssh and ftp.

All I want is the ftp back.

Whoops! I just took a look at the proftpd.conf file. It changed on the day I was screwing up, but I know I didn't touch it. In fact, I never touched it. You install proftpd and it just works. But here's how the file starts:

ServerType standalone
DefaultServer on
Umask 022
ServerName "0.0.0.0"
ServerIdent on "My FTPD"
ServerAdmin [email]Admin@this.domain.topd[/email]omain
IdentLookups off
UseReverseDNS off
Port 21
PassivePorts 49152 65534
#MasqueradeAddress None
TimesGMT off

As I understand it, ServerType should be static, no? Wish the system would make backups when it changes files without telling you.

Some other stuff looks fishy to me. Any ideas?

[GreyGeek]

Re: Messed up my "network"

GreyGeek, I've probably bogged you down with way more information than you need, and confused things in the process. For that I apologize.

In a nutshell:

1. My serial modem works fine, and connects to my ISP just fine. It's plugged into the serial port, as it should be.

2. The only time I would put up eth0 is to connect my 2 boxes via the crossover cable, so that I can transfer files from one to the other.

3. The boxes can ping each other.

O.K. I think I get the picture now. When you fire up your Hardy box you can connect to your ISP via your dial up and Kppp, and resolv.conf is fine. It is when you try to make a connection to your Jaunty box that you break Hardy's resolv.conf and you still don't have ftp connectivity to Jaunty.


Are you trying to set up a dhcp connection to Jaunty via the eth card in Hardy (the dhcpcd daemon running on Jaunty supposedly responding to the dhcp client running on Hardy)? Attempting to do that is what is overwriting your ISP DNS info in resolv.conf with the loopback address of Jaunty, i.e., Jaunty's "DNS". Unfortunately it is also the loopback address of Hardy too, so both boxes terminate their signals in the 127.0.0.1 blackhole.

Manual method:

For Hardy:
sudo ifconfig eth0 192.168.1.2 netmask 255.255.255.0

For Jaunty:
sudo ifconfig eth0 192.168.1.3 netmask 255.255.255.0

Default gateway for both: 192.168.1.1

(don't use 0 or 254 or 255 as the last quad in any address. Those are special.

GUI method:

Install wicd on both Hardy and Jaunty. Open wicd chose STATIC IP addresses for your two eth cards, giving them the settings above. You don't need the dhcp or dhcpcd.

[nodough]

Re: Messed up my "network"

Sorry for the delay in replying. Life got in the way.

O.K. I think I get the picture now. When you fire up your Hardy box you can connect to your ISP via your dial up and Kppp, and resolv.conf is fine. It is when you try to make a connection to your Jaunty box that you break Hardy's resolv.conf and you still don't have ftp connectivity to Jaunty.

Almost

Just looked at my resolv.conf again before I went online. It was fine, with the correct nameservers.

So I go online with my modem, and now it says nameserver 127.0.0.1 again.

And don't ask me how it works with nameserver 127.0.0.1, but it's definitely use of the serial modem that's changing it, because I haven't been able to use the ethernet card.

Are you trying to set up a dhcp connection to Jaunty via the eth card in Hardy

At this point, I have no idea what I'm trying to do, except get things back to the way they were To my knowledge, I wasn't using dhcp before.

Manual method:

For Hardy:
sudo ifconfig eth0 192.168.1.2 netmask 255.255.255.0

For Jaunty:
sudo ifconfig eth0 192.168.1.3 netmask 255.255.255.0

Default gateway for both: 192.168.1.1

I set them up through the system settings, but I presume it doesn't make any difference?

Sigh! Hardy is still refusing connections, but they're still pinging.

I went through my bash history and found this:
sudo apt-get install dnsmasq iptables
sudo apt-get install resolvconf
sudo iptables -t nat -L

And within the resolvconf directory there's a file /run/interface/lo.dnsmasq which reads

nameserver 127.0.0.1

Geez, what would happen if I just deleted that? Or better, renamed it. Yes, I know I could try it on my own, but with files changing on me without notice, I'm afraid to try anything anymore.

Install wicd on both Hardy and Jaunty.

Before I do any installing on Hardy, I have to ask:
Is there any way in the world wicd could affect my serial modem? Because if I lose my internet connection, I am in a huge load of trouble.

My other option is to install proftpd on Jaunty and see if that works. It won't fix the problem, but it's a workaround and I can at least transfer files.

Thanks muchly for your patience.

[oshunluvr]

Re: Messed up my "network"

will the ethernet cards connect manually

startup booth machines, connect to the internet with your modem, then open a terminal on both computers.

Type one of these lines in each one:

ifconfig eth0 192.168.1.2 netmask 255.255.255.0 up
ifconfig eth0 192.168.1.3 netmask 255.255.255.0 up

and then see if you can ping and see your internet connection remains active.

Also - have you checked your route table?

On the computer with the modem - before you dial-up, in a terminal type:

route -n

then connect and do it again, I'm guessing the results are different.

Also - check the file (or whatever ppp uses these days for dialup config) /etc/ppp/ip-up.d or possibly under /etc/default/ppp

One last idea - look at /etc/dhcp3/dhclient.conf I think it should look something like

prepend domain-name-servers 127.0.0.1;

Sorry I'm all over the place, just trying to give some other places to look - and I haven't used ppp for quite a while.

[nodough]

Re: Messed up my "network"

will the ethernet cards connect manually

startup booth machines, connect to the internet with your modem, then open a terminal on both computers.

Type one of these lines in each one:

ifconfig eth0 192.168.1.2 netmask 255.255.255.0 up
ifconfig eth0 192.168.1.3 netmask 255.255.255.0 up

and then see if you can ping and see your internet connection remains active.

I wasn't entirely sure how specifically you meant for me to follow this. I didn't reboot the machines, if that's what you intended. Also, I just used sudo ifconfig eth0 up to get the ethernet card going.

However, after that:

1. They ping
2. My internet connection still works.

Also - have you checked your route table?

Not connected to internet: (eth0 was up)

route -n
Kernel IP routing table
Destination   Gateway     Genmask     Flags Metric Ref  Use Iface
192.168.0.0   0.0.0.0     255.255.255.0  U   0   0    0 eth0

Connected to internet: (eth0 was down)

route -n
Kernel IP routing table
Destination   Gateway     Genmask     Flags Metric Ref  Use Iface
134.117.137.5  0.0.0.0     255.255.255.255 UH  0   0    0 ppp0
0.0.0.0     0.0.0.0     0.0.0.0     U   0   0    0 ppp0

Connected to internet: (eth0 was up)

route -n
Kernel IP routing table
Destination   Gateway     Genmask     Flags Metric Ref  Use Iface
134.117.137.3  0.0.0.0     255.255.255.255 UH  0   0    0 ppp0
192.168.0.0   0.0.0.0     255.255.255.0  U   0   0    0 eth0
0.0.0.0     0.0.0.0     0.0.0.0     U   0   0    0 ppp0

Also - check the file (or whatever ppp uses these days for dialup config) /etc/ppp/ip-up.d or possibly under /etc/default/ppp

All looks okay to me. Of course, I never checked these before the mess-up.

One last idea - look at /etc/dhcp3/dhclient.conf I think it should look something like

prepend domain-name-servers 127.0.0.1;

It did originally. Now the name servers are my ISP's DNS numbers. I did that manually, though (I think).

Sorry I'm all over the place, just trying to give some other places to look - and I haven't used ppp for quite a while.

Understood. And much appreciated, because I can't figure out WHAT is going on.

[GreyGeek]

Re: Messed up my "network"

Sorry for the delay in replying. Life got in the way.

O.K. I think I get the picture now. When you fire up your Hardy box you can connect to your ISP via your dial up and Kppp, and resolv.conf is fine. It is when you try to make a connection to your Jaunty box that you break Hardy's resolv.conf and you still don't have ftp connectivity to Jaunty.

Almost

Just looked at my resolv.conf again before I went online. It was fine, with the correct nameservers.

So I go online with my modem, and now it says nameserver 127.0.0.1 again.

And don't ask me how it works with nameserver 127.0.0.1, but it's definitely use of the serial modem that's changing it, because I haven't been able to use the ethernet card.

Ah! You explain it yourself, down below.

Are you trying to set up a dhcp connection to Jaunty via the eth card in Hardy

At this point, I have no idea what I'm trying to do, except get things back to the way they were To my knowledge, I wasn't using dhcp before.

Manual method:

For Hardy:
sudo ifconfig eth0 192.168.1.2 netmask 255.255.255.0

For Jaunty:
sudo ifconfig eth0 192.168.1.3 netmask 255.255.255.0

Default gateway for both: 192.168.1.1

I set them up through the system settings, but I presume it doesn't make any difference?

No. Folks who don't like the CLI find the GUI easier.

Sigh! Hardy is still refusing connections, but they're still pinging.

I went through my bash history and found this:
sudo apt-get install dnsmasq iptables
sudo apt-get install resolvconf
sudo iptables -t nat -L

And within the resolvconf directory there's a file /run/interface/lo.dnsmasq which reads

nameserver 127.0.0.1

Geez, what would happen if I just deleted that? Or better, renamed it. Yes, I know I could try it on my own, but with files changing on me without notice, I'm afraid to try anything anymore.

All of those files are preinstalled, at least on my Lucid. You just reinstalled them, possibly overwriting default configuration files. Your last command, "sudo aptables -t net -L" appears to have added to the damage by attempting to set up masquerading. .

Here is a comment in the dnsmasq setup file:

Automatic DNS server configuration with PPP.
Later versions of pppd have an option "usepeerdns" which instructs it to write a file containing the address(es) of the DNS severs in /etc/ppp/resolv.conf. Configure dnsmasq as above with "nameserver 127.0.0.1" in /etc/resolv.conf and run dnsmasq with to option resolv-file=/etc/ppp/resolv.conf.

The ONLY way your FireFox can still be finding domain names when /etc/resolv.conf contains "127.0.0.1" is if your Kppp program created /etc/ppp/resolv.conf, and dnsmasq is using it. Check for the existance of /etc/ppp/resolv.conf and if you find it tell us what it contains.

(K)Ubuntu already has a firewall, ufw, which configures the already installed iptables. the ufw has Its own gui, gufw, which you can install with Synaptic. And its an easy way to turn iptables on or off. I suspect that if you check the "Enable" check box, then immediately uncheck it to turn off the default iptable rules, your should be able to browse with FF.

Install wicd on both Hardy and Jaunty.

Before I do any installing on Hardy, I have to ask:
Is there any way in the world wicd could affect my serial modem? Because if I lose my Internet connection, I am in a huge load of trouble.

While I have dnsmasq & iptables & resolvconf installed, but I have not run or configured any of them. As long as you can connect with Kppp stick with it.

My other option is to install proftpd on Jaunty and see if that works. It won't fix the problem, but it's a workaround and I can at least transfer files.

Thanks muchly for your patience.

Knowing how to back all your changes out, and guiding you via this thread to recover, is probably beyond my patience and OUR abilities. It would be a LOT quicker to use the LiveCD you installed your distro with and reinstall it. Then, just use Kppp to make your connection to the Internet, and install wicd and use it to connect the eth card in hardy with the eth card in jaunty. Don't mess with dnsmasq or iptables configuration files.

[nodough]

Re: Messed up my "network"

Hi GreyGeek:

I think we both reached the same stage of burnout at the same time.

Folks who don't like the CLI find the GUI easier.

Having looked at it in Jaunty, I just got confused.

It would be a LOT quicker to use the LiveCD you installed your distro with and reinstall it.

Ah, but try reinstalling all those programs on dialup

What I did was install proftpd on Jaunty, and the boxes are connecting that way.

All of those files are preinstalled, at least on my Lucid. You just reinstalled them, possibly overwriting default configuration files.

Not on Hardy. However, believe it or not, I think you've just solved the problem. My /etc/network/interfaces file has been rewritten! It's got some lines in it it didn't have before, including one for eth1.

The panic of getting files from a possibly wonky hard drive backed up to the Jaunty box has been mitigated, so I may now play around with that /etc/network/interfaces.

I thank you hugely for your time and patience. I really did learn a lot--especially that I don't know as much as I thought I did (which was precious little).

Thanks, too, to oshunluvr for his contribution.

[oshunluvr]

Re: Messed up my "network"

Yokes - thank goodness we got it fixed! Whew...