Second attempt to install OpenBSD on my troublesome system

[aleunix]

How i can be 100 % sure that my computer work before the installation?
I tried NetBSD 5.1 (livecd) and the keyboard stop to function when i'm on the prompt. So i can't use it at all.
Yesterday i tried even OpenBSD 5.0 (livecd) and seem ok.
Which information - and using which command from the livecd - i must give you for be 100 % sure that everything work even after a real installation?
If everything is ok i will remove one linux distro from my system so i can give an all disk at openbsd (6gb of an old disk ...).

[jggimi]

If you used one of my 5.0 live media images, the OS is 5.0-release. The only difference is a modified set of rc(8) scripts.

[aleunix]

Yes i used your livecd.
Now I've found that using the livecd I can't access the internet how to fix?

[jggimi]

Is your network connection WiFi?

It was recently brought to my attention that because WiFi connections are not automated in my scripts, this may be an issue for those new to the OS. Wired Ethernet connections are automated, as long as the connection is physically made before booting.

To enable a WiFi, DHCP connection:

1. Discover the name of the attached WiFi NIC driver:

$ ifconfig

This will list all of the network interfaces, including pseudo devices like lo0 and pflog0. If you have both wired and wireless NICs, look for a NIC with an associated "nwid" field -- the SSID.

2. Define your SSID to attach to, and associated security key if used:

Open network:
# ifconfig <nic> nwid <SSID>

WPA/WPA2 with pre-shared key:
# ifconfig <nic> nwid <SSID> wpakey <passphrase>

WEP:
# ifconfig <nic> nwid <SSID> nwkey <key>

3. Request a TCP/IP configuration from the DHCP server on your LAN:

# dhclient <nic>

----

For more information, please refer to OpenBSD FAQ 6 and these man pages: ifconfig(8) and dhclient(8).

[aleunix]

Originally Posted by jggimi

Is your network connection WiFi?

No.
It's a normal Wired Ethernet connections to my router.

[jggimi]

During boot my script will search through the NICs looking for connected networks, and will run dhclient to attempt to obtain a TCP/IP configuration.

Possible causes for not having a valid network connection after boot:

a. Not physically connecting the LAN before booting
b. Not having a DHCP server on the LAN
c. Not having a supported NIC.

You should know if a. or b. are the reason for the trouble. As for c. ...

Check the output of "ifconfig" - the pseudo NICs will be lo0, pflog0, and enc0. Do you have any other NICs shown?

Check your dmesg. Look for devices which are recognized, but for which there is no driver. They are shown as "not configured". Any NICs without drivers?

$ dmesg | grep 'not configured'

[aleunix]

No issue using static or dinamic IP with Linux.

For type the command more easy i tried to change the layout of the keyboard
#wsconsctl -w keyboard.encoding=it
but seem without success.

I will try to write on paper what you ask.

[jggimi]

You can route the output of those commands to filesystems on USB or diskette, if that is easier than writing them out by hand. See OpenBSD FAQ 4.16 for an example of capturing a dmesg from the bsd.rd ramdisk installer, which would be similar.

[jggimi]

Mounting these filesystems if defined via MBR is fairly easy. Step 1 - use disklabel(8) to determine what partition letters have been assigned to the partitions. IDE drives are wd0, wd1, wd2... and SCSI/USB/SATA drives are sd0, sd1, sd2... For example:

# disklabel sd2

That will produce a report of the partition letters assigned by OpenBSD (virtually) for the third SCSI/USB/SATA drive. Foreign filesystems with MBR partitions will be assigned letters between "i" and "p". Partition "c" is reserved, and is always the entire drive.

Step 2:

Mount the desired partition. Example:

# mount /dev/sd2i /mnt

That will mount the "i" partition of drive sd2 as /mnt.

[jggimi]

One more comment:

Foreign filesystems that may be mounted read write are: FAT (12/16/32), EXT2, and EXT3 (as EXT2). These will appear in disklabel text as MSDOS and ext2fs.

[aleunix]

Using ifconfig i have:
lo0
rc0: ethernet autoselect (10baseT half-duplex)
enc0
pflog0

$ dmesg | grep 'not configured'
"VIA VT6306 Firewire ..."
no other message.

On linux, using:
$ ip addr

1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
inet 127.0.0.1/8 scope host lo
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast state UP qlen 1000
link/ether 00:1b:fc:6f:ec:69 brd ff:ff:ff:ff:ff:ff
inet 192.168.1.65/24 brd 192.168.1.255 scope global eth0
inet6 fe80::21b:fcff:fe6f:ec69/64 scope link
valid_lft forever preferred_lft forever

How i can use the same IP on the livecd?
Perhaps if is change the dhcp is possible resolve more fast this issue.

I already tried:
#route add -net 162.168.1.65/24 192.168.1.1
but without success.

Note:
I made this new restart before you posted these new message about the use of disklabel.

[aleunix]

Now i have make another attempt using disklabel.
So, finally, i can give you the data as seen by openbsd 5.

$ dmesg | grep 'not configured'
"VIA VT6306 FireWire" rev 0xc0 at pci3 dev 7 function 0 not configured
I don't use it.

[jggimi]

Excellent! Thank you!
Here is what I see, from ifconfig:

re0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
    lladdr 00:1b:fc:6f:ec:69
    priority: 0
    media: Ethernet autoselect (10baseT half-duplex)
    status: no carrier

The NIC and its PHY are supported, and the specific chipsets are noted in the dmesg:

re0 at pci2 dev 0 function 0 "Realtek 8168" rev 0x01: RTL8168 2 (0x3800), apic 2 int 19, address 00:1b:fc:6f:ec:69
rgephy0 at re0 phy 7: RTL8169S/8110S PHY, rev. 2

This gives the appearance of a cable connection problem -- your 100Mbps NIC shows that it is training at 10Mb half duplex, without being able to connect to the Ethernet.

[aleunix]

But if this was a cable connection problem, this system should not work using any other OS.
So i think is a driver problem.

If is available i can even try a current livecd (but i don't find it) on the hope that a more update driver can solve that problem.

[jggimi]

I agree with you, this is a problem that appears to be software.

I do not have -current live media ... but the OpenBSD Project's -current install media can be used for testing.

1. Download /pub/OpenBSD/snapshots/<arch>/cd51.iso from your nearest mirror, burn to CD, and boot.

2. At the "(i)nstall, (u)pgrade, or (s)hell" prompt, type "s" and press Enter.

3. # ifconfig re0

---

If the output from ifconfig looks the same, then the problem still exists. If, however,
you now show a connected Ethernet (and perhaps better than 10mbps/half-duplex), you may check to see if you can obtain a network connection:

4. # dhclient re0

-current is now about a week or two in development beyond what will be 5.1-release, expected to be available on or about 1 May 2012.

[aleunix]

Now i tried it but i obtain the same output.
Anyway there is one thing that i don't understand, if i made, for example:
#ping www.google.it

PING www-cctld.l.google.com (173.194.35.56): 56 data bytes
64 bytes from 173.194.35.56: icmp_seq=0 ttl=58 time=36.010 ms
64 bytes from 173.194.35.56: icmp_seq=1 ttl=58 time=36.511 ms
64 bytes from 173.194.35.56: icmp_seq=2 ttl=58 time=35.202 ms
64 bytes from 173.194.35.56: icmp_seq=3 ttl=58 time=34.885 ms

--- www-cctld.l.google.com ping statistics ---
4 packets transmitted, 4 packets received, 0.0% packet loss
round-trip min/avg/max/std-dev = 34.885/35.652/36.511/0.643 ms

or
#ping www.yahoo.it

PING any-rc.a01.yahoodns.net (98.139.102.145): 56 data bytes
64 bytes from 98.139.102.145: icmp_seq=0 ttl=46 time=141.113 ms
64 bytes from 98.139.102.145: icmp_seq=1 ttl=46 time=144.782 ms
64 bytes from 98.139.102.145: icmp_seq=2 ttl=46 time=147.668 ms
64 bytes from 98.139.102.145: icmp_seq=3 ttl=46 time=147.342 ms
64 bytes from 98.139.102.145: icmp_seq=4 ttl=46 time=141.848 ms

--- any-rc.a01.yahoodns.net ping statistics ---
5 packets transmitted, 5 packets received, 0.0% packet loss
round-trip min/avg/max/std-dev = 141.113/144.550/147.668/2.740 ms

How is possible?
What i mean, if the connection don't work, i should obtain 0 packets received and 100 % packet loss and don't this output.

[jggimi]

Your IPv4 TCP/IP configuration comes from DHCP -- if ping is working, you must now have a TCP/IP configuration, and the output from ifconfig must be different for re0.

You cannot still be showing "no carrier" and have no IPv4 configuration, and have a working IPv4 ping function as you've shown.

[aleunix]

#ifconfig -a

lo0: flags=8008<LOOPBACK,MULTICAST> mtu 33152
re0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
lladdr 00:1b:fc:6f:ec:69
media: Ethernet autoselect (10baseT half-duplex)
status: unknown

If you want, next time, i copy even the output for:
# ifconfig re0

[jggimi]

I believe you. But I am very confused, because, in order for an IPv4 ping by domain name to function, the following must be in place:

1. A valid domain name resolver configuration must be in /etc/resolv.conf
2. A valid IPv4 TCP/IP configuration must be established, including IP address, subnet mask, and default route.
3. ARP traffic must flow on the local Ethernet to establish a relationship between MAC addresses and IP addresses.
4. UDP traffic must flow to the domain name server addressed to destination port 53, and a response received.
5. ICMP traffic must flow to the destination IP address.

#1 and #2 would normally occur through DHCP. From what you have posted, it appears that #2 did not take place. Without #2.... #3,#4, and #5 would seem to be impossible.

After a "successful" ping such as you have shown, without a valid IP configuration, I would want to know:

• Is there an /etc/resolv.conf? What are its contents?
• Are there a list of MAC to IP addresses from "arp -an" ?? Is the re0 mac address (00:1b:fc:6f:ec:69) listed?
• What does the IPv4 routing table look like? What is the output of "route -n show -inet" ?

I have never seen "status: unknown" with ifconfig output before. I am going to guess that there is a communication error between the NIC and PHY drivers.

[aleunix]

I finally found the issue!
If, after the shell, i take the output from the command
# ifconfig re0
before run
# dhclient re0
the network give that strange result while if i take it after the run of the dhclient, i obtain the correct output.
So I asked myself, why not do a new test with the livecd 5 - i used the fluxbox image - but, this time, forcing the launch of the dhclient after i already inside of fluxbox, despite, this time, everything should be automatic.
And it is work!
So, in the end, the issue seem inside the livecd, where strangely the dhclient seem don't be launched during the boot.

$ ifconfig re0

re0: flags=8802<BROADCAST,SIMPLEX,MULTICAST> mtu 1500
lladdr 00:1b:fc:6f:ec:69
priority: 0
media: Ethernet autoselect (10baseT half-duplex)
status: no carrier

# dhclient re0

DHCPDISCOVER on re0 to 255.255.255.255 port 67 interval 1
...

# ifconfig re0

re0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
lladdr 00:1b:fc:6f:ec:69
priority: 0
groups: egress
media: Ethernet autoselect (100baseTX full-duplex,rxpause,txpause)
status: active
inet6 fe80::21b:fcff:fe6f:ec69%re0 prefixlen 64 scopeid 0x1
inet xxxx netmask 0xffffff00 broadcast xxxxxx

Note
Now i' m writing from the inside of the livecd openbsd 5.
Using the follow command from inside the livecd
# wsconsctl keyboard.encoding=it
keyboard.encoding -> it
don' t give the change so was more difficult write the things, i have made some copy and past on some letters.