On Thu, 4 Oct 2007, Collins, Kevin [MindWorks] wrote:
Started with virt-manager and installed with kickstart. The kickstart
config is generated by a script, and the only difference for these hosts
are things pertaining to host name and IP address.
and earlier
in the last few days I have patched 6 paravirtualized hosts
running RHEL5. All of these hosts were built identically,
ummm -- How? manually built with virt-manager, or scripted
built with virt-install? 'built identically' is so general ;)
No solution, but a workaround:
For the last week, we have been repeatedly kickstart building
and testing instances of the same image, over and over, and
have not seen _that_ issue.
We did this our builds, using first, virt-manager; wanting to
lose the requirement for the GUI, we then more recently
shifted to a virt-install approach, in the same fashion
kickstart, and seem to have uncovered a failure to honor the
'-p' virt-install option, when passed down into a sub process,
involving Intel processors, with PAE but not VMX, with SELinux
Enabled/Targeted. I have a reproducer script and a
cooperative pastebin trace, and need to file it yet.
When we moved over to the virt-install approach, I added a
%post to our ks.php delivery URL, emitting this:
[herrold ~]$ lynx -dump http://10.16.1.231/pub/kickstart/ks.php
...
%post
# add a console for Xen connections
cat - << END >> /etc/inittab
co:2345:respawn:/sbin/agetty xvc0 9600 vt100-nav
END
# add a console for Xen connections
cat - << END >> /etc/securetty
xvc0
END
....
--------------------------
That HERE document logic can be tweaked obviously to test if
the installer has already attended to the issue you are
seeing, and to optionally skip repetition. It turns out that
duplicated lines in the relevant inittab and securetty are
simply noted and ignored, and ignored, respectively, so I did
not bother.
-- Russ Herrold
_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list
