This is a bug. I think I never tested a dirinstall with -N, I've
always used -c. I did some changes in setting/reading the hostname in
FAI 5.1 and 5.1.2, so this may be the cause of your problems.
Can you please file a bug report for this? I'm on vacation, so not
much time to look deeper into it n
Hello everybody,
bringing debian-lan up-to-date for stretch, I run into a bug/problem.
A dirinstall with the command:
fai -vNu diskless dirinstall $DLROOT
ends up using not the class 'diskless' but the hostname of the
faiserver, in my case 'mainserver':
root@mainserver:~# cat /var/log/fa
Thomas,
We just discovered the same thing here. I actually changed the PXE boot
file for the client to use NFSv3 (by adding ':vers=3' to the end of the
'root=' line), but the effect was the same. I think using NFSv3 for the
installs will be an OK workaround for now. Thanks for you help with thi
Hi Jeff, hi Stephen,
I found the problem. It's NFS v4 again. If I export my nfsroot via
NFS v4 I also have the 'operation not supported' error message when
doing a touch /tmp/test.
On your FAI install server you can add this line to the /etc/exports
/srv/nfs4 1.2.3.4/25(fsid=0,ro,no_subtre
Hi there,
I'm trying to install a stretch VM with more or less the same settings I
have for jessie (same config directory).
Almost everything works quite fine but grub is not able to find my root
on LVM as it was before.
#disk_config:
> disk_config disk1 disklabel:msdos bootable:1
> primary -
Jeff, (btw, i work with Jeff)
I think it's useful to add some of the other stuff we discovered here.
The overlay fs seems to be setup in 'dracut', and not having ever used
this, i was confused by the parameters for upperdir+workdir
(/cow/{rw,work}) not existing in the running fai environment.
appa
Thomas,
Thank you for creating the nfsroot tarball. Unfortunately, using this
nfsroot produces the same problem for me (/tmp not writeable). If
this nfsroot works OK for you, then my problem must be somewhere else.
Maybe a boot parameter or NFS export option? I'll keep looking and let
you know
On Tue, Aug 9, 2016, at 07:38 PM, andrew bezella wrote:
> On Tue, 2016-08-09 at 14:12 -0700, Ian Kelling wrote:
> >
> > Yes, I'm following this path. At the top of 30-interface, I have a
> > condition
> > for if the os is in a class that uses a persistent name, and inside
> > I have this code:
>