Hi Roberto,
On 8/22/19 12:34 AM, Roberto Ibarra Magdaleno wrote:
Tried this configuration:
root=/dev/ram0 ro ip=off ramdisk_size=40000
rescue
and still booting installer, not a rescue system:
Starting sshd to allow login over the
network.
Connect now to 172.27.20.19 and log in as user install to start the
installation
.
E.g. using: ssh -x install@172.27.20.19
Those steps might still all be necessary for the rescue option kicking in
later, see further down.
You may log in as the root user to start an interactive shell.
This is also a possible alternative, but depending on which phases/stages the
install process has dynamically loaded into the ramdisk, the available tools
can be very minimal.
(You can even get a root shell on the console (without network) just be
pressing enter (twice in a z/VM guest).
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/installation_guide/ch-s390-phase_1#ch-s390-Phase_1-terminals)
In the installer root ssh I have no commands for LVM:
I looked at the content of RHEL6.10 images/initrd.img and it is indeed very
minimal.
Welcome to the anaconda install environment 1.2 for zSeries
[anaconda root@linux8 root]# lsdasd
Bus-ID Status Name Device Type BlkSz Size Blocks
==============================================================================
0.0.0200 active dasdb 94:4 ECKD ??? 2347MB ???
0.0.0201 active dasdc 94:8 ECKD ??? 2347MB ???
0.0.0202 active dasdd 94:12 ECKD ??? 2347MB ???
0.0.0203 active dasde 94:16 ECKD ??? 2347MB ???
0.0.0204 active dasdf 94:20 ECKD ??? 7043MB ???
0.0.0205 active dasdg 94:24 ECKD ??? 7043MB ???
0.0.0206 active dasdh 94:28 ECKD ??? 7043MB ???
So at least the 2 new minidisks are there and active, that's good.
Anaconda and its environment tooling handle cio_ignore transparently for the
user, so usually there is no need to change the cio_ignore= kernel parameter.
[anaconda root@linux8 root]# pvdisplay
-bash: pvdisplay: command not found
See further down for a possibly different syntax due to ramdisk space reasons,
not necessarily providing all the individual end user process binary names for
the LVM tooling suite.
You’re right those are PVs in an LVM, maybe I would try to activate the
vg_root in another Linux (called RESCUE by the way) but wouldn’t It cause
problems with the running root? Since the rescue RHEL system (that I can’t
start) mounts it as /mnt/sysimage.
From an LVM assembly point of view it should not cause problems as the VGs
should have different names.
I suppose you would mount the LVM to be fixed under some free mount point and
chroot there.
Still going to try from the Linux “RESCUE” virtual machine.
That's a good alternative.
whereas RHEL6 has an older slightly different syntax:
IBM Z:
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/installation_guide/ch-parmfiles-miscellaneous_parameters
general (not everything applies to Z):
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/installation_guide/ap-rescuemode#s1-rescuemode-boot
Does this help?
I think you don't even need the cms conf file for the rescue system, as
the
latter runs off the initrd and without network. The cms conf file is only
parsed by the very early installer phase which should not run in rescue
mode.
I guess I stand corrected. My apologies. The "rescue" option is part of
anaconda and that lives in install.img which is in turn loaded by initrd.img,
typically over the network on s390.
So probably you do need all the installer network setup on the kernel parm file
including an indirect pointer to the install.img by means of repo=
[https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/installation_guide/ch-parmfiles-loader_parameters].
I think repo= in the parm file is optional. If you run without it, connect as
install user over ssh, then loader runs and you can teach the repo location
interactively in the TUI
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/installation_guide/s1-installationmethod-s390.
IIRC, the rescue will only start after above and the step where it pulls
install.img
https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/installation_guide/ch22s06
Only then you would get the rescue part instead of the anaconda installer.
PC users booting the installer of a DVD don't notice that as the boot process
would automatically load initrd.img which would automatically load install.img
from the same DVD and so the "rescue" boot option simply materializes without
further steps there. PC users installing over the network would be more like
the process on s390.
CMSDASD=191 CMSCONFFILE=redhat.conf
REDHAT CONF A1
DASD="200-206"
HOSTNAME="linux8"
NETTYPE="qeth"
IPADDR="172.27.20.19"
SUBCHANNELS="0.0.0900,0.0.0901,0.0.0902"
NETMASK="255.255.0.0"
GATEWAY="172.27.20.254"
Basically, in the dracut (initrd) rescue shell you can manually prepare
all
necessary dependency (devices) for the root-fs. Then try to exit the
rescue
shell and it will try (again) to mount the root-fs and continue to boot.
https://mirrors.edge.kernel.org/pub/linux/utils/boot/dracut/dracut.html#accessing-the-root-volume-from-the-dracut-shell
(step 2 sounds quite like your use case with root-fs on LVM)
# lvm vgscan
# lvm vgchange -ay
--
Mit freundlichen Gruessen / Kind regards
Steffen Maier
Linux on IBM Z Development
https://www.ibm.com/privacy/us/en/
IBM Deutschland Research & Development GmbH
Vorsitzender des Aufsichtsrats: Matthias Hartmann
Geschaeftsfuehrung: Dirk Wittkopp
Sitz der Gesellschaft: Boeblingen
Registergericht: Amtsgericht Stuttgart, HRB 243294
----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit
http://www2.marist.edu/htbin/wlvindex?LINUX-390
