The difference between dev-branch and master branch appears to be http vs https.
Here’s the commit that made the change:
https://github.com/FOGProject/fogproject/commit/349a8fa2df00d82df857527b1cf45aa204363834
Introduced in 1.5.9.58 of dev-branch
Today pictured below, for example. This is what normally happens. Master fails and dev-branch succeeds. Yet, dev-branch and master branch are executing against the same OS, same patch level. The way these tests are setup is - the VMs get a snapshot restored. Then they are patched. Then they get the specified branch installed. This is repeated for each branch. Fresh snapshot applied and fresh patching performed before every installation.
CentOS 8 Output for master branch: 2021-07-30_12-00-PM_output.log
CentOS 8 Error Log for master branch: 2021-07-30_12-00-PM_fog_error.log
dev-branch fog_error output for CentOS 8:
package remi-release-8* is not installed
warning: /var/tmp/rpm-tmp.AVYE4j: Header V4 RSA/SHA256 Signature, key ID 5f11735a: NOKEY
Retrieving https://rpms.remirepo.net/enterprise/remi-release-8.rpm
Verifying... ########################################
Preparing... ########################################
Updating / installing...
remi-release-8.4-1.el8.remi ########################################
Extra Packages for Enterprise Linux Modular 8 - 1.9 MB/s | 798 kB 00:00
Extra Packages for Enterprise Linux 8 - x86_64 24 MB/s | 10 MB 00:00
Remi's Modular repository for Enterprise Linux 772 B/s | 858 B 00:01
Remi's Modular repository for Enterprise Linux 3.0 MB/s | 3.1 kB 00:00
Importing GPG key 0x5F11735A:
Userid : "Remi's RPM repository <remi@remirepo.net>"
Fingerprint: 6B38 FEA7 231F 87F5 2B9C A9D8 5550 9759 5F11 735A
From : /etc/pki/rpm-gpg/RPM-GPG-KEY-remi.el8
Remi's Modular repository for Enterprise Linux 795 kB/s | 779 kB 00:00
Safe Remi's RPM repository for Enterprise Linux 1.6 kB/s | 858 B 00:00
Safe Remi's RPM repository for Enterprise Linux 3.0 MB/s | 3.1 kB 00:00
Importing GPG key 0x5F11735A:
Userid : "Remi's RPM repository <remi@remirepo.net>"
Fingerprint: 6B38 FEA7 231F 87F5 2B9C A9D8 5550 9759 5F11 735A
From : /etc/pki/rpm-gpg/RPM-GPG-KEY-remi.el8
Safe Remi's RPM repository for Enterprise Linux 1.9 MB/s | 1.9 MB 00:01
bc-1.07.1-5.el8.x86_64
curl-7.61.1-18.el8.x86_64
master branch fog_error output for CentOS8:
package remi-release-8* is not installed
warning: /var/tmp/rpm-tmp.fFeNbj: Header V4 RSA/SHA256 Signature, key ID 5f11735a: NOKEY
Retrieving http://rpms.remirepo.net/enterprise/remi-release-8.rpm
Verifying... ########################################
Preparing... ########################################
Updating / installing...
remi-release-8.4-1.el8.remi ########################################
Extra Packages for Enterprise Linux Modular 8 - 4.7 MB/s | 798 kB 00:00
Extra Packages for Enterprise Linux 8 - x86_64 32 MB/s | 10 MB 00:00
Remi's Modular repository for Enterprise Linux 568 B/s | 858 B 00:01
Remi's Modular repository for Enterprise Linux 3.0 MB/s | 3.1 kB 00:00
Importing GPG key 0x5F11735A:
Userid : "Remi's RPM repository <remi@remirepo.net>"
Fingerprint: 6B38 FEA7 231F 87F5 2B9C A9D8 5550 9759 5F11 735A
From : /etc/pki/rpm-gpg/RPM-GPG-KEY-remi.el8
Is this ok [y/N]: Remi's Modular repository for Enterprise Linux 2.4 kB/s | 858 B 00:00
Error: Failed to download metadata for repo 'remi-modular': repomd.xml GPG signature verification error: Bad GPG signature
bc-1.07.1-5.el8.x86_64
curl-7.61.1-18.el8.x86_64
@sebastian-roth Yeah, today’s tests look like a mess. Though, what your saying about this GPG key. All the tests use the same OS, same patch level each day. If there was an issue with a GPG key in master branch, wouldn’t it be true that the same problem exists in dev branch? Most days, this is not the case. Most days, just master branch fail for CentOS 8 and RHEL 8.
There are ways to inject commands before installation, though the point of these tests is just to see in an automatic way if there are any issues that arise for a typical installation.
@sebastian-roth said in RHEL 8, CentOS 8, master branch, July 22, 2021:
Seems like the REMI GPG key changed. You need to manually run dnf update once and confirm the GPG key import.
Interesting this is not happening for the dev-branch tests.
@Sebastian-Roth I have done this. Which is why I’m bringing it up again.
I’ve been monitoring & have once cleaned up the daily installation tests, and for a while now RHEL 8 and CentOS 8 have been failing every day on the master branch, while the dev branch passes tests every day.
Given the dev branch is working fine, it would seem the fix is to release the current dev as a new version. What sort of shape is the dev branch concerning all the other aspects of FOG? Is it in a shape to release?
The external reporting graphs show 50 servers on 1.5.9.98, with overall about 460 servers operating on the dev branch.
There is another aspect to this - maybe we don’t prioritize a release just for fixing CentOS 8 and RHEL8. It’s a legitimate question to ask, at least. External reporting shows CentOS 8 and CentOS Stream 8 being around 15 total for the dev branch. Pretty low figure. RHEL doesn’t even register in the top 20 OS versions in use. Raspbian has a higher adoption on dev branch than RHEL.
The Chrome browser will offer to translate for you usually. Of course, I don’t endorse the Chrome web browser. Just mentioning it’s capabilities.
If you’ve sys-prep’d the system correctly and the image has drivers for the VM, then it should work fine. The easy route is just to build the image you want inside the VM, sysprep it, and capture it.
@tom-elliott in that case, I would agree that putting the extras into the SANs is appropriate. I’ll see if I can figure this out.
@tom-elliott if that were implemented, then in @xburnerx00’s case his certificate would have an IP that the FOG server isn’t using after one reboot.
In what legitimate case does a single Linux interface have two IPs?
@xburnerx00 said in New install failing:
I changed the settings to manual, as it was already set to DHCP.
I’ve seen this behavior lots of times with Linux. If you do a DHCP release / renew with these commands:
dhclient -r && dhclient
it produces the same behavior. the old address is held on the interface along with the new one. I’ve found that rebooting clears it. I’ve not tried other things.
Reference: https://linux.die.net/man/8/dhclient
I’ll work on a check for detecting two IPs on the chosen interface. If there are two, the installer should note this and exit rather than falling on it’s face and people not knowing what’s going on. It can also offer advice on how to fix (like rebooting)
Ubuntu 16.04 is now officially “End of Standard Support”.
https://wiki.ubuntu.com/Releases
This means packages in “main” and “restricted” are no longer supported for it.
@sebastian-roth I know, need to fix it.
This happens every time there is a GRUB update for Ubuntu. Even though I specify apt-get -y dist-upgrade it asks what to do, every time, and leaves the run in a bad state.
@sebastian-roth We should have renamed that user donotusethisuserever
@tesparza You may want to look through the location plugin documentation. Most of your questions I think will be answered by this and the video in there. https://wiki.fogproject.org/wiki/index.php?title=Location_Plugin
Wanting to open a discussion about Ansible. The FOG installation scripts are quite complex and large. They do a good job, have for years. Though I wonder if Ansible scripts would be easier to maintain? This is something I might try to write.
@sebastian-roth said in FOG Project call for engagement:
Interesting we don’t see any 21.04 in the install reporting yet. At least two have posted in the forums so far.
For the exact OSs graph, only the top 20 are shown. I can change it to show more, but the graph bars will start getting really thin.
I can step up on the installer scripts. That’s where my skills most align. I’ve dabbled in this area a lot already, and have the automated tests around this.
@sebastian-roth Great job, the installation tests are looking beautiful today.
I would suggest sticking to the Ubuntu LTS versions for use with FOG. I test the latest LTS versions daily (results in my signature). It may be that this version of PHP isn’t available for Ubuntu 21.04. Also, to update FOG you really need to have internet access too.