A root-cert in for Let's encrypt expired on this date:
https://letsencrypt.org/docs/dst-root-ca-x3-expiration-september-2021
Our jobs in gerrit installing machines, for example matrix-runner, for
stretch and buster, fail to download packages from debian. and
deb.sipwise.com complaining that the certificate is invalid.
This is because of having older versions of ca-certificates, that don't
contain the new root-cert, or because versions of some libraries
(e.g. gnutls used by wget and apt) do not seem to work well when there
are several chains to validate the certificate -- they use only the
first chain, and if elements of it are invalid/expired, don't try the
others.
This change forces installation of new versions of ca-certificates and
libgnutls30 both in the "outside" system running grml (because it needs
to use "wget" accessing the servers for special commands) as well as
"inside" the new chroot, because by default in the new Debian system
ca-certificates is not installed.
Change-Id: I58ff9daa10c21f5ad867308c33ffa303c120683d
(cherry picked from commit 79b0f26a7b,
adapted due to many conflicts backporting)
(cherry picked from commit 1bdf053f6c)
grub-pc >=2.04-11 has a new behavior regarding /boot/grub/i386-pc/
handling, where we end up with an empty /boot/grub/i386-pc/ after
*successful* grub-install execution:
| root@grml ~ # vgchange -ay
| 3 logical volume(s) in volume group "ngcp" now active
| root@grml ~ # mount /dev/mapper/ngcp-root /mnt
| root@grml ~ # grml-chroot /mnt /bin/bash
| Writing /etc/debian_chroot ...
| (spce)root@grml:/# cd
| (spce)root@grml:~# grub-install /dev/sda
| Installing for i386-pc platform.
| Installation finished. No error reported.
| (spce)root@grml:~# ls -la /boot/grub/i386-pc/
| total 16
| drwxr-xr-x 2 root root 12288 Dec 16 12:04 .
| drwxr-xr-x 4 root root 4096 Dec 16 12:07 ..
This causes the installed system to fail to boot with:
| GRUB loading..
| Welcome to GRUB!
|
| error: file `/boot/grub/i386-pc/normal.mod' not found.
| grub rescue> _
The underlying issue is that recent grub versions unlink the files
inside /boot/grub/i386-pc, though it doesn't report anything about it
(even under `--verbose` execution).
This is triggered in our situation, as lvm2's vgs binary isn't present
yet. In earlier versions of grub this wasn't causing any problems and
grub-install happily installed the files inside /boot/grub/i386-pc, even
though we installed lvm2 only afterwards via our metapackages. To
ensure lvm2 is available during installation time within
grml-debootstrap, explicitly add to it list of packages to be installed.
See https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=977544 for further
details regarding the grub bug.
Change-Id: I27a1cd18777526eb26b838fae88d4d87b6e93467
(cherry picked from commit 91e047a486)
* Addon fixes:
* Provide custom addon template according to our needs, instead of
relying on the default one being present in grml-live.git/templates
* Install memtest86+ as addon for BIOS usage
* Install netboot.xyz.lkrn for BIOS usage and netboot.xyz.efi for EFI usage
* Install ipxe.lkrn for BIOS usage and ipxe.efi for EFI usage
* Stick grml-live version to v0.34.3, instead of relying on some random
git master version
* Stick grml2usb version to v0.17.0, instead of relying on the grml2usb
version available on the host system (being 0.14.14 on Debian/stretch as
present on our current build hosts). For arbitrary addon file names we
need grml2iso (which uses grml2usb underneath) from grml2usb >=0.17.0.
FTR, grml2iso and grml2usb can be executed from within the git repository,
assuming all relevant tools are present
* No longer invoke isohybrid on the resulting ISO, instead rely on
grml2iso behaviour (which also checks for EFI support and enables
according switches as needed)
* Fix usage instructions in t/Dockerfile:
* it's "deployment-iso-buster" and not "lua-ngcp-kamailio-buster"
* refer to working directory instead of "deployment-iso.git",
which very probably isn't named as such on any of our systems,
while the $(pwd) approach works for c/p
* Fix docker build usage in grml_build/Dockerfile (for building we need
to provide a PATH (being current working directory for us)
* Provide testing tools in grml-build-buster docker environment
* Provide new testing script t/iso-tester to compare generated ISO
against pre-defined screenshot (only testing memtest feature using
./t/screenshots/01-memtest.jpg for now)
Change-Id: I67e3f85bbe86bd1b3ee709161504b5250ca5d7fe
(cherry picked from commit c3d753562d)
In mr7.5+ bootstrap gpg keyring is named sipwise-keyring-bootstrap.gpg
so add appropriate directory.
Change-Id: I633deac1ffb203e7d566e5262e0dff35354aa2e5
This file contains the following public keys:
68A702B1FD8E422AAAA1ADA3773236EFF411A836 - jessie, stretch, buster
debian repos, mr3.8-mr8.4 public repos.
F7B8A739CE638D719A078C9859104633EE5E097D - autobuild repos.
8A61E3C73B6987020226079D69D9C21F6D9B587E - new key for mr7.5+.
This very file will be used only for bootstrap purposes and will be
dropped in the end of installer.
Change-Id: I60a903b4b05ab8a0d1a6498773802f756f8e4202
Initially dirmngr package was installed to allow running apt-key inside
deployment.sh to install puppetlabs key. Then after moving puppetlabs
key to local file package installation was removed. But we still need it
for successful first puppet run that also uses apt-key. Otherwise there
is the error: 'Could not find a suitable provider for apt_key'. Puppet
also installs dirmngr package but in such case only the second puppet run
will be successful so we need to install the package before the very
first puppet run.
Change-Id: I28edbd8c91c841074ac1b3f1eb6df16e14daa084
In deployment.sh line 2109:
local puppet_gpg='/root/puppet.gpg'
^-- SC2168: 'local' is only valid in functions.
Change-Id: I628cbac844db6aa1913ab1747906656d7d6d739b
The puppet.gpg is located in Grml-sipwise so no need to download the
hard-coded key but use the local one.
Change-Id: Id57180de96efef4a5a13086e7807c615bd65b886
In the event of network blips, HA switchovers or failure of the servers, there
might be a problem downloading some of the hundreds or thousands of packages
needed for installation.
Set the option to retry 3 times in the case of such failures, to try to minimise
their impact.
This has to be done in several places because mmdebstrap (which uses apt) and
apt-get are invoked at different points.
Change-Id: I5acbd9895fa37452026e582c241ced945fbba3d7
This script copies puppet.gpg file to '/root' dir of Grml-sipwise
image in building process.
Create 'scripts/PUPPETLABS' so '10-gpgkey' is copied there in runtime.
Change-Id: I836fa35e3f64f40cb4ee4a298fc18676f7689b54
The boot parameter 'ngcpstatus' is used to wait the number of seconds
before ending process so the external system can retrieve the final
status from 4242 port.
Add the value of this parameter as STATUS_WAIT_SECONDS option to
ngcp-initial-configuration config so external system can track the
configuration stage also.
Change-Id: I68d9c65a1cc96bf581305e92d8b9b6bfc34e7ed2
mmdebstrap and debootstrap install any "important" packages in the
standard installation variants, pulling in e.g. libgdbm5 (which was of
priority `important` until #890832 was fixed and it was downgraded to
priority `optional` with libgdbm6).
Same for libusb-1.0-0 vs libusb-0.1-4:amd64:
% apt-cache show libusb-1.0-0 | grep Priority
Priority: optional
% apt-cache show libusb-0.1-4:amd64 | grep Priority
Priority: important
Resulting in having libusb-0.1-4 *and* libusb-1.0-0 installed on
our NGCP systems, while libusb-0.1-4 is completely unneeded and irrelevant.
So this is unnecessary, and all the other packages that are automatically
pulled in via the standard installation should be handled via according
and explicit dependencies elsewhere, JFTR - currently being in
Debian/buster:
| apt-utils bsdmainutils cpio cron debconf-i18n dmidecode dmsetup
| e2fsprogs gdbm-l10n ifupdown init iproute2 iptables iputils-ping
| isc-dhcp-client isc-dhcp-common kmod less libapparmor1 libapt-inst2.0
| libargon2-1 libbsd0 libcap2 libcap2-bin libcom-err2 libcryptsetup12
| libdevmapper1.02.1 libdns-export1104 libelf1 libestr0 libext2fs2
| libfastjson4 libidn11 libip4tc0 libip6tc0 libiptc0 libisc-export1100
| libjson-c3 libkmod2 liblocale-gettext-perl liblognorm5 libmnl0
| libncurses6 libnetfilter-conntrack3 libnewt0.52 libnfnetlink0 libnftnl11
| libpopt0 libprocps7 libslang2 libss2 libtext-charwidth-perl
| libtext-iconv-perl libtext-wrapi18n-perl libxtables12 logrotate lsb-base
| mount nano netbase procps readline-common rsyslog sensible-utils systemd
| systemd-sysv tasksel tasksel-data tzdata udev vim-common vim-tiny
| whiptail xxd
The essential variant lacks apt(-get), while the minbase variant provides
the same base as essential, *plus* apt so we can actually install
further packages, so that's what we're using.
We pull in systemd + systemd-sysv + init explicitly to force its usage,
ngcp-installer is complaining and prompting otherwise.
We pull in isc-dhcp-client + ifupdown to have a working networking/DHCP
setup after reboot, since we don't use systemd-networkd or alike (yet).
Applying this change will result in lacking the following packages in
our Debian/stretch based trunk builds:
* apt-utils
* blends-tasks
* debconf-i18n
* iputils-ping
* isc-dhcp-common
* libapt-inst2.0
* liblocale-gettext-perl
* libtext-charwidth-perl
* libtext-iconv-perl
* libtext-wrapi18n-perl
* libusb-0.1-4
* logrotate
* tasksel
* tasksel-data
* vim-tiny
The ones we want to have available will be handled via explicit
dependencies in NGCP (meta)packages.
Change-Id: I14dac92d99172cf792a0334601a930ce6698dc83
Sipwise DNS servers in LDC will be turned off soonish,
also new DNS servers in GCloud will NOT be available for public usage.
We have to use some public DNS servers.
Change-Id: I1c5cb78b90da18e893658417a6886b7b4f81fa0d
This variable tells ngcp-installer that it is run from
deployment.sh.
In this case we need it to skip CE warning in ngcp-installer:
===================
This installation script is not intended to run in a shared system,
as it can add/delete/update existing configurations.
Please run this script only in a base install of 64 bit Debian 9 (stretch)
on a dedicated server.
Do you want to continue with the installation process? (y/N):
===================
Change-Id: I9a33ccbe07332a09a64c98a2bef844fa95498c05
By default we should not skip any confirmation messages. In automation
setup we need just change it to 'yes' before
ngcp-initial-configuration.
Change-Id: I1ca69aaa1e19f8e34b82434d99b50e41add74f6f
Sipwise Jenkins Builder
Manuel Montecelo
Michael Prokop
Mykola Malkov
Alexander Lutay
Guillem Jover
Sergii Kipot