|
See additions to my previous reply
See my reply interspersed. (from 08/20/2018 06:48 PM)
On 08/20/2018 10:26 AM, Dusty Mabe
wrote:
On 08/19/2018 10:50 AM, Bob Gustafson wrote:
My particular piece of bare metal seems to have enough memory, disk
space, Intel x86_64 architecture w/4 cores, ethernet nics, but it does
not have any video support.
I can move data using ethernet, but user interaction during install must
go through a 115200n8 serial port.
Do you see the isolinux boot menu on the serial console? If so you just need
to edit the first entry and add console=ttyS0 to the kernel command line.
Thanks very much for your reply.
I don't get that far on the serial console. I think that
dialog is being painted to a video frame buffer..
This is what I see on an attempt using a USB stick:
SeaBIOS (version
rel-1.10.0.1)
Press F10 key now for boot menu, N for PXE
boot
Select boot
device:
1. USB MSC Drive PMAP << I chose this
item
2. SD card SS04G
3781MiB
3. ata0-0: Samsung SSD 850 EVO mSATA 250GB ATA-9 Hard-Disk
(23
4. iPXE (PCI
00:00.0)
5. Payload
[memtest]
6. Payload
[setup]
Booting from Hard Disk...
Blinking USB light, also attempts by anaconda to feel around
its environment
From Wireshark:
..
Option: (60) Vendor class identifier
Length: 44
Vendor class identifier: anaconda-Linux
4.16.3-301.fc28.x86_64 x86_64
I have avoided using Anaconda since F20, but for this first install, I
think interacting with Anaconda would be the best route.
(I anticipate installing many times as I feel out the bare metal and
increase my own knowledge.)
If you are going to install many many times I would suggest using something
like PXE and a kickstart based install.
I have tried many different approaches to installation - I
favor paths with visibility. I turned on logging from the
dnsmasq server and am running Wireshark on every attempt.
Kickstart seems to favor a non-interactive series of steps and
so config errors are not so catchable.
Using a USB stick - I see the little red light blink many
times with quiet periods in between, then dark. Nothing on the
console after the "Booting from Hard Disk..." line. I'm sure
that I am only a "misplaced comma" away from success, but ..
Currently I am watching Wireshark list continuing DHCP
Discover packets, but no reply from the dnsmasq server..
Perhaps the line in systemctl status dnsmasq -> "ignoring
nameserver 192.168.50.60 - local interface" has some
relationship
No, the ignoring nameserver line only pertains to the DNS part
of dnsmasq. The non-reply from dnsmasq was caused by no DHCP
server running, thus no reply.
The dnsmasq config seems to *require* a dhcp-range statement.
A single dhcp-host statement does not bring up the DHCP server
component - even though I want an ip address for only one dhcp
requesting device.
A modified dnsmasq config contains the following active
statements:
[root hoho8 dnsmasq.d]# sed -e '/^#/d' dnsmasq-edited.conf
| grep -v "^$"
no-poll
server=/49.168.192.in-addr.arpa/192.168.49.41 # far
away on eno1
server=/50.168.192.in-addr.arpa/192.168.50.60 # on
enp3s0
bind-dynamic # maybe need this because enp3s0 goes up
and down ?
interface=enp3s0
no-hosts
dhcp-range=192.168.50.53,192.168.50.60,255.255.255.0,12h
# proper range needed or no-go?
dhcp-host=id:00:0d:b9:46:55:78,192.168.50.55
dhcp-host=id:00:1b:21:4c:f0:98,192.168.50.60
dhcp-match=set:efi-x86_64,option:client-arch,7
dhcp-match=set:efi-x86_64,option:client-arch,9
dhcp-match=set:efi-x86,option:client-arch,6
dhcp-match=set:bios,option:client-arch,0
dhcp-boot=tag:efi-x86_64,"efi64/syslinux.efi"
dhcp-boot=tag:efi-x86,"efi32/syslinux.efi"
dhcp-boot=tag:bios,"pxelinux/lpxelinux.0"
enable-tftp
tftp-root=/var/lib/tftpboot
log-queries
log-dhcp
[root hoho8 dnsmasq.d]#
This config file results in the following running listeners:
[root hoho8 dnsmasq.d]# netstat -nlp | grep dnsmasq | grep
udp
udp 0 0 192.168.50.60:53
0.0.0.0:* 17623/dnsmasq
udp 0 0 127.0.0.1:53
0.0.0.0:* 17623/dnsmasq
udp 0 0 0.0.0.0:67
0.0.0.0:* 17623/dnsmasq
udp 0 0 192.168.50.60:69
0.0.0.0:* 17623/dnsmasq
udp 0 0 127.0.0.1:69
0.0.0.0:* 17623/dnsmasq
udp6 0 0 ::1:53
:::* 17623/dnsmasq
udp6 0 0 ::1:69
:::* 17623/dnsmasq
[root hoho8 dnsmasq.d]#
Note the line containing "0.0.0.0:67" This is the dhcp server
needed to service the DHCP Discover packet.
That mole whacked, now on to tweeking the console=ttyS0
somewhere.
Just one more whack-a-mole...
-----------------
Kickstart is probably my eventual installation technique,
especially for the 2nd and 3rd piece of bare metal. But for
the 1st, there are still too many pieces of the puzzle that
are out of place. There is the dnsmasq configuration, a
combination of dhcp, dns, tfptd. Fortunately most of these
commas are visible through Wireshark. Many of these pieces I
haven't had a previous need to conquer, and it has been slow
slogging.
Fedora Documentation with working examples would be a big
help. (but everyone wants 'their use case' as an example..)
I have installed a working Linux system (Voyage) on my bare
metal, but of course that is not available during the pre-boot
and boot phases of Fedora Atomic.
My host system is a healthy Fedora 28. Minicom is my serial port
software - seems to work ok.
The Fedora-AtomicHost-ostree-x86_64-28-20180806.0.iso has vesamenu.c32,
but no menu.c32 file.
I'm looking for suggestions as to how I should proceed.
See my reply interspersed.
On 08/20/2018 10:26 AM, Dusty Mabe
wrote:
On 08/19/2018 10:50 AM, Bob Gustafson wrote:
My particular piece of bare metal seems to have enough memory, disk
space, Intel x86_64 architecture w/4 cores, ethernet nics, but it does
not have any video support.
I can move data using ethernet, but user interaction during install must
go through a 115200n8 serial port.
Do you see the isolinux boot menu on the serial console? If so you just need
to edit the first entry and add console=ttyS0 to the kernel command line.
Thanks very much for your reply.
I don't get that far on the serial console. I think that dialog is
being painted to a video frame buffer..
This is what I see on an attempt using a USB stick:
SeaBIOS (version
rel-1.10.0.1)
Press F10 key now for boot menu, N for PXE
boot
Select boot
device:
1. USB MSC Drive PMAP << I chose this item
2. SD card SS04G
3781MiB
3. ata0-0: Samsung SSD 850 EVO mSATA 250GB ATA-9 Hard-Disk
(23
4. iPXE (PCI
00:00.0)
5. Payload
[memtest]
6. Payload
[setup]
Booting from Hard Disk...
Blinking USB light, also attempts by anaconda to feel around its
environment
From Wireshark:
..
Option: (60) Vendor class identifier
Length: 44
Vendor class identifier: anaconda-Linux 4.16.3-301.fc28.x86_64
x86_64
I have avoided using Anaconda since F20, but for this first install, I
think interacting with Anaconda would be the best route.
(I anticipate installing many times as I feel out the bare metal and
increase my own knowledge.)
If you are going to install many many times I would suggest using something
like PXE and a kickstart based install.
I have tried many different approaches to installation - I favor
paths with visibility. I turned on logging from the dnsmasq server
and am running Wireshark on every attempt. Kickstart seems to
favor a non-interactive series of steps and so config errors are
not so catchable.
Using a USB stick - I see the little red light blink many times
with quiet periods in between, then dark. Nothing on the console
after the "Booting from Hard Disk..." line. I'm sure that I am
only a "misplaced comma" away from success, but ..
Currently I am watching Wireshark list continuing DHCP Discover
packets, but no reply from the dnsmasq server..
Perhaps the line in systemctl status dnsmasq -> "ignoring
nameserver 192.168.50.60 - local interface" has some relationship
Just one more whack-a-mole...
-----------------
Kickstart is probably my eventual installation technique,
especially for the 2nd and 3rd piece of bare metal. But for the
1st, there are still too many pieces of the puzzle that are out of
place. There is the dnsmasq configuration, a combination of dhcp,
dns, tfptd. Fortunately most of these commas are visible through
Wireshark. Many of these pieces I haven't had a previous need to
conquer, and it has been slow slogging.
Fedora Documentation with working examples would be a big help.
(but everyone wants 'their use case' as an example..)
I have installed a working Linux system (Voyage) on my bare metal,
but of course that is not available during the pre-boot and boot
phases of Fedora Atomic.
My host system is a healthy Fedora 28. Minicom is my serial port
software - seems to work ok.
The Fedora-AtomicHost-ostree-x86_64-28-20180806.0.iso has vesamenu.c32,
but no menu.c32 file.
I'm looking for suggestions as to how I should proceed.
|