From Fedora Project Wiki

No edit summary
 
Line 909: Line 909:


   
   
[[Category : Fedora 36 Test Day]]
[[Category:Fedora 36 Test Days]]

Latest revision as of 10:09, 28 April 2023

Fedora CoreOS Test Week
Echo-testing-48px.png
Fedora CoreOS

Date 2022-04-04 to 2022-04-11
Time all week

Website QA/Test Days
IRC #fedora-coreos (webirc)
Mailing list test


Note.png
Can't make the date?
If you come to this page before or after the test day is completed, your testing is still valuable, and you can use the information on this page to test, file any bugs you find at Fedora CoreOS issue tracker, and add your results to the results section. If this page is more than a month old when you arrive here, please check the current schedule and see if a similar but more recent Test Day is planned or has already happened.

What to test?[edit]

Today's installment of Fedora Test Day will focus on Fedora CoreOS

Who's available?[edit]

The following cast of characters will be available for testing, workarounds, bug fixes, and general discussion ...

For real time help, please join us on the IRC: #fedora-coreos[?] on https://libera.chat/.

Documentation is also available here. Documentation feedback is welcome through chat, mailing list, github tracker and in the form of a pull request to the documentation sources.

Prerequisites for Test Day[edit]

  • Virtual machine (x86_64)
  • Test day Image

Grab images/artifacts/information for the most current next stream release (36) from our download page: https://getfedora.org/en/coreos/download?tab=cloud_operators&stream=next

How to test?[edit]

Run the tests[edit]

Visit the results page and click on the column title links to see the tests that need to be run: most column titles are links to a specific test case. Follow the instructions there, then enter your results by clicking the Enter result button for the test.

Reporting bugs[edit]

Please report all bugs, issues or enhancement proposals to the Fedora CoreOS issue tracker.

If you are unsure about exactly how to file the report or what other information to include, just ask on IRC #fedora-coreos[?], #fedora-test-day[?] or #fedora-qa[?] and we will help you.

Test Results[edit]

Installation[edit]

User Profile Virtual install Bare Metal install References
Daniels QEMU - KVM (fedora-coreos-36.20220410.1.1-live.x86_64.iso) Host: Gnome 41.5 Intel® Pentium(R) CPU 4415U @ 2.30GHz × 4 Lenovo 2in1 14
Pass pass
[1]
  1. ssh ((port (core@localhost) access denied to 127.0.0.1:2222. coreos 410)) Earlier versions of Coreos : 2022313.325. 327: Virtual Install. (No Pass) librvirt test Installs (OS: fedora-coreos-next, not recognized ) Smooth testing otherwise.
Nemric PXE installation
Pass pass
[1]
Pass pass
[2]
  1. Configured swap on Zram with default settings
  2. Running : 1) docker.io/grafana/promtail:latest instance 2) quay.io/prometheus/node-exporter:latest instance 3) docker.io/library/registry:latest and docker.io/library/redis:latest in a pod
andilinux
Pass pass
andilinux GNOME BOXES
Pass pass
[1]
  1. Fedora CoreOS next all worked fine for me
bittin FCOS next image
Pass pass
bittin FCOS next image in Virtualbox
Pass pass
Fail fail
checking FCOS next image
Pass pass
danniel FCOS next image
Pass pass
donaldsebleung FCOS next image with QEMU; arch: x86_64; custom Ignition file with password-based SSH access; launching with QEMU
Warning warn
[1]
Pass pass
[2]
  1. Can confirm that IP address of VM is printed in FCOS stable but not FCOS next, using identical configuration. Does not seem to affect usability of system though
  2. Using ignition file: https://github.com/DonaldKellett/fcos-test-day-cloud-launch/blob/main/fcos-test-day.ign ; following documentation "launching with QEMU": https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-qemu/ ; did not see IP address of VM printed to console possibly because password-based SSH was configured instead of SSH key(s) ; both local and SSH password login worked as expected, did not find any system issues
dustymabe Installed FCOS next on Intel NUC
Pass pass
ersen virt-manager
Pass pass
geraldosimiao FCOS next image
Pass pass
[1]
  1. tested qemu qcow and iso image too
kisk fedora-coreos-36.20220325.1.0-qemu.x86_64.qcow2.xz - libvirt
Pass pass
lnie fcos next image,x86_64,VM created by testcloud
Pass pass
lruzicka QEMU
Pass pass
[1]
  1. I was able to install using the approach described in https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-qemu/, however the approach described here https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-libvirt/ did not work and the command was complaining about no permission to read the ignition file which was the same (with same permissions) in both attempts.
millerthegorilla
Warning warn
[1]
Fail fail
[2]
Fail fail
[3]
Fail fail
[4]
Warning warn
[5]
Fail fail
[6]
  1. these were both the x86_64 image
  2. mounted virtual image as per https://gist.github.com/shamil/62935d9b456a6f9877b5 (Mount qcow image), and installed using https://docs.fedoraproject.org/en-US/fedora-coreos/bare-metal/#_installing_from_the_container. Installer ran fine, sshd in, and no ip address was reported on screen. got the ip address from virt-manager and ssh'd in. Seems to be working, tried machinectl shell to open standard user from ignition file, worked fine.
  3. started live cd install using virt-manager. Started ok, to prompt, but when trying to run coreos-installer picking up config.ign from the local network, the install failed with 'error checking whether /dev/sda is an ibm dasd disk.' Output of lsblk is garbled, with disk listed as '^%\xef\x9aaU'. Perhaps related to this (https://discussion.fedoraproject.org/t/coreos-pxeboot-coreos-installer-cant-format-dev-vda-or-dev-sda/25680)
  4. edk2 install failed to present a network address to the internet, wired connection not recognised by router.
  5. Docs for uboot at (https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-raspberry-pi4/#_installing_fcos_and_booting_via_u_boot) assume a non rpm-ostree install on the host machine. Ie there is no dnf available on my silverblue laptop, and I couldn't figure how to install using toolbox (perhaps I should have used a --privileged flag to podman). The docs for EDK (https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-raspberry-pi4/#_edk2_combined_fedora_coreos_edk2_firmware_disk), should probably invite the user to expand their xfs partition before mounting. In my experience, only gparted has the capacity to do this correctly without too much hassle.
  6. install on pi4b following instructions for edk2 at [fedora-docs](https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-raspberry-pi4/#_edk2_combined_disk_mode_alternate_machine_disk_preparation) - boot loads, after very long pause at grub boot screen 'booting coreos-vernum' , No IP address is shown, and SSH host key is displayed, 3 times, one for ecdsa, one for ed25519 and one for RSA. What are these host keys, and why are they displayed?
pnemade FCOS next image
Pass pass
sayaksarkar GNOME Boxes
Pass pass
sumantrom QEMU
Pass pass

Cloud launch[edit]

User Profile AWS Azure GCP DigitalOcean VMWare Exoscale IBM Cloud VirtualBox Vultr OpenStack Alibaba References
alciregi DigitalOcean droplet
Pass pass
[1]
Warning warn
[2]
  1. ok, works (in the documentation: in doctl compute image create the "--wait" option thorws an error)
  2. Droplet creation successfully. Can't ssh: I don't know if it is a problem on my side (ssh key, ignition file) or what else. Will try again later
danniel FCOS next image
Pass pass
donaldsebleung Fedora CoreOS 36 Next stream (arch: x86_64; ami: ami-066680d6ebe90fce1) on AWS
Pass pass
[1]
donaldsebleung Fedora CoreOS 36 Next stream (arch: x86_64; instance type: e2-micro) on GCP
Pass pass
[1]
dustymabe 36.20220410.1.1 on s-2vcpu-2gb droplet in nyc3
Pass pass
dustymabe exoscale
Fail fail
[1]
dustymabe ibmcloud bx2-2x8 instance
Pass pass
fifofonix
Pass pass
[1]
  1. Successfully created and ssh'd into a fedora:36 next machine on vSphere 6.7 using govc following instructions linked from the test case. Some issues following govc instructions and suggested documentation improvements here: https://github.com/coreos/fedora-coreos-tracker/issues/1158
jdoss
Pass pass
[1]
  1. Fedora CoreOS 36.20220325.1.0 launched just fine on GCP with a e2-medium instance in us-central1-a
miabbott FCOS 36 on Aliyun
Pass pass
[1]
  1. Used fedora-coreos-36.20220325.1.0-aliyun.x86_64.qcow2
mnguyen
Pass pass
[1]
Pass pass
[2]
  1. Booted FCOS 36.20220325.1.0 on Azure in centralus following https://fedoraproject.org/wiki/QA:Testcase_CoreOS_Azure
  2. Provisioned FCOS 36.20220325.1.0 on OpenStack following https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-openstack/. I did not need to assign floating IP for my OpenStack instance so those steps were not followed.
ravanelli Fedora CoreOS 36 Next stream (arch: x86_64) on VirtualBox
Pass pass
[1]
  1. 1. Successfully booted fedora-coreos-36.20220325.1.0-live.x86_64.iso on VirtualBox using these instructions: https://fedoraproject.org/wiki/QA:Testcase_CoreOS_Provisioning_on_VirtualBox
sayaksarkar fedora-coreos-36.20220325.1.0-live.x86_64.iso
Pass pass
sohank2602 Fedora CoreOS 36 Next stream (arch: x86_64) on Vultr
Pass pass
[1]
  1. Successfully booted fedora-coreos-36.20220325.1.0-vultr.x86_64 on Vultr using these instructions: https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-vultr/

aarch64[edit]

User Profile AWS Virtual install Bare Metal install References
Daniels QEMU - KVM (fedora-coreos-36.20220410.1.1-live.x86_64.iso) Host: Gnome 41.5 Intel® Pentium(R) CPU 4415U @ 2.30GHz × 4 Lenovo 2in1 14
Pass pass
[1]
  1. ssh ((port (core@localhost) access denied to 127.0.0.1:2222. coreos 410)) Earlier versions of Coreos : 2022313.325. 327: Virtual Install. (No Pass) librvirt test Installs (OS: fedora-coreos-next, not recognized ) Smooth testing otherwise.
Nemric PXE installation
Pass pass
[1]
Pass pass
[2]
  1. Configured swap on Zram with default settings
  2. Running : 1) docker.io/grafana/promtail:latest instance 2) quay.io/prometheus/node-exporter:latest instance 3) docker.io/library/registry:latest and docker.io/library/redis:latest in a pod
andilinux
Pass pass
andilinux GNOME BOXES
Pass pass
[1]
  1. Fedora CoreOS next all worked fine for me
bittin FCOS next image
Pass pass
bittin FCOS next image in Virtualbox
Pass pass
Fail fail
checking FCOS next image
Pass pass
danniel FCOS next image
Pass pass
donaldsebleung FCOS next image with QEMU; arch: x86_64; custom Ignition file with password-based SSH access; launching with QEMU
Warning warn
[1]
Pass pass
[2]
  1. Can confirm that IP address of VM is printed in FCOS stable but not FCOS next, using identical configuration. Does not seem to affect usability of system though
  2. Using ignition file: https://github.com/DonaldKellett/fcos-test-day-cloud-launch/blob/main/fcos-test-day.ign ; following documentation "launching with QEMU": https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-qemu/ ; did not see IP address of VM printed to console possibly because password-based SSH was configured instead of SSH key(s) ; both local and SSH password login worked as expected, did not find any system issues
donaldsebleung Fedora CoreOS 36 Next stream (arch: x86_64; ami: ami-066680d6ebe90fce1) on AWS
Pass pass
[1]
dustymabe Installed FCOS next on Intel NUC
Pass pass
ersen virt-manager
Pass pass
geraldosimiao FCOS next image
Pass pass
[1]
  1. tested qemu qcow and iso image too
kisk fedora-coreos-36.20220325.1.0-qemu.x86_64.qcow2.xz - libvirt
Pass pass
lnie fcos next image,x86_64,VM created by testcloud
Pass pass
lruzicka QEMU
Pass pass
[1]
  1. I was able to install using the approach described in https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-qemu/, however the approach described here https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-libvirt/ did not work and the command was complaining about no permission to read the ignition file which was the same (with same permissions) in both attempts.
millerthegorilla
Warning warn
[1]
Fail fail
[2]
Fail fail
[3]
Fail fail
[4]
Warning warn
[5]
Fail fail
[6]
  1. these were both the x86_64 image
  2. mounted virtual image as per https://gist.github.com/shamil/62935d9b456a6f9877b5 (Mount qcow image), and installed using https://docs.fedoraproject.org/en-US/fedora-coreos/bare-metal/#_installing_from_the_container. Installer ran fine, sshd in, and no ip address was reported on screen. got the ip address from virt-manager and ssh'd in. Seems to be working, tried machinectl shell to open standard user from ignition file, worked fine.
  3. started live cd install using virt-manager. Started ok, to prompt, but when trying to run coreos-installer picking up config.ign from the local network, the install failed with 'error checking whether /dev/sda is an ibm dasd disk.' Output of lsblk is garbled, with disk listed as '^%\xef\x9aaU'. Perhaps related to this (https://discussion.fedoraproject.org/t/coreos-pxeboot-coreos-installer-cant-format-dev-vda-or-dev-sda/25680)
  4. edk2 install failed to present a network address to the internet, wired connection not recognised by router.
  5. Docs for uboot at (https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-raspberry-pi4/#_installing_fcos_and_booting_via_u_boot) assume a non rpm-ostree install on the host machine. Ie there is no dnf available on my silverblue laptop, and I couldn't figure how to install using toolbox (perhaps I should have used a --privileged flag to podman). The docs for EDK (https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-raspberry-pi4/#_edk2_combined_fedora_coreos_edk2_firmware_disk), should probably invite the user to expand their xfs partition before mounting. In my experience, only gparted has the capacity to do this correctly without too much hassle.
  6. install on pi4b following instructions for edk2 at [fedora-docs](https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-raspberry-pi4/#_edk2_combined_disk_mode_alternate_machine_disk_preparation) - boot loads, after very long pause at grub boot screen 'booting coreos-vernum' , No IP address is shown, and SSH host key is displayed, 3 times, one for ecdsa, one for ed25519 and one for RSA. What are these host keys, and why are they displayed?
pnemade FCOS next image
Pass pass
sayaksarkar GNOME Boxes
Pass pass
sumantrom QEMU
Pass pass

Advanced configuration[edit]

User Profile Static networking Complex partitioning Building containers Containerized service Kernel Tuning (sysctl) Modifying Kernel Arguments OS extensions Configuring SwapOnZRAM Configuring Time Zone Debugging with Toolbox Customizing NIC name Setting alternatives Node counting References
Nemric FCOS updated from "next 35" to "next 36" on baremetal
Pass pass
[1]
Pass pass
  1. Running : - docker.io/grafana/promtail:latest instance - quay.io/prometheus/node-exporter:latest instance - docker.io/library/registry:latest and docker.io/library/redis:latest in a pod Services were running well before and after running 'podman system reset' -> https://docs.fedoraproject.org/en-US/fedora-coreos/major-changes/#_podman_v4_0
Nemric PXE installation
Pass pass
Pass pass
aaradhak
Pass pass
[1]
Pass pass
  1. Executed some debugging commands using toolbox version : registry.fedoraproject.org/fedora-toolbox:35
aaradhak FCOS 36 next stream
Pass pass
[1]
  1. Debugged toolbox with strace command and it was succcessful
alciregi DigitalOcean droplet
Pass pass
Pass pass
Fail fail
[1]
Pass pass
Pass pass
Pass pass
Pass pass
  1. Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: Problem: conflicting requests Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: - package vim-enhanced-2:8.2.4701-1.fc36.x86_64 requires vim-common = 2:8.2.4701-1.fc36, but none of the providers can> Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: - package vim-enhanced-2:8.2.4621-1.fc36.x86_64 requires vim-common = 2:8.2.4621-1.fc36, but none of the providers can> Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: - package vim-common-2:8.2.4701-1.fc36.x86_64 requires vim-data = 2:8.2.4701-1.fc36, but none of the providers can be > Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: - package vim-common-2:8.2.4621-1.fc36.x86_64 requires vim-data = 2:8.2.4621-1.fc36, but none of the providers can be > Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: - cannot install both vim-data-2:8.2.4701-1.fc36.noarch and vim-data-2:8.2.4579-1.fc36.noarch Apr 10 09:27:49 my-fcos-droplet rpm-ostree[1434]: - cannot install both vim-data-2:8.2.4621-1.fc36.noarch and vim-data-2:8.2.4579-1.fc36.noarch Apr 10 09:27:49 my-fcos-droplet systemd[1]: rpm-ostree-install-vim.service: Main process exited, code=exited, status=1/FAILURE
checking FCOS next image
Pass pass
donaldsebleung FCOS 36.20220325.1.0 on QEMU/KVM
Warning warn
[1]
Pass pass
[2]
Pass pass
[3]
Pass pass
[4]
Pass pass
[5]
Pass pass
[6]
Pass pass
[7]
Fail fail
[8]
  1. Did not see IP address of VM printed to console after boot (prints when using "stable" stream), but does not affect usability of system
  2. Static networking worked as expected. Configured IP address and netmask 10.0.2.10/24, gateway 10.0.2.2. ip addr show gives correct IP, pinging the VM, host and external websites all work as expected
  3. Test case "Adding /var/log from a secondary disk" works as expected: second disk is detected, partition created and formatted with xfs filesystem as specified, filesystem mounted on /var/log as specified
  4. Rootless podman, privileged podman, privileged docker all passed
  5. Busybox hello.service works as expected
  6. kernel.sysrq test works as expected, custom value of 0 remains after a reboot
  7. Setting systemd.unified_cgroup_hierarchy=0 in the Ignition config worked as expected: got a visible warning message on SSH login, /proc/cmdline indicates the kernel argument has been set. Rebooted the VM, settings persisted. Used rpm-ostree kargs to (1) update systemd.unified_cgroup_hierarchy to 1; (2) remove systemd.unified_cgroup_hierarchy; (3) add systemd.unified_cgroup_hierarchy=0. All changes were correctly registered after a reboot.
  8. Following the example "Layering vim and setting it as the default editor" in the docs results in a failed service with the following messages: Problem: conflicting requests - package vim-enhanced-2:8.2.4621-1.fc36.x86_64 requires vim-common = 2:8.2.4621-1.fc36, but none of the providers can be installed - package vim-common-2:8.2.4621-1.fc36.x86_64 requires vim-data = 2:8.2.4621-1.fc36, but none of the providers can be installed - cannot install both vim-data-2:8.2.4621-1.fc36.noarch and vim-data-2:8.2.4579-1.fc36.noarch . EDITOR is set correctly to vim though. Also the docs should mention that this service might take a while to run before the reboot.
hricky Intel® Core™ i7-4790 CPU, 32 GB RAM, Bare Metal
Warning warn
[1]
  1. Building and running container using podman in rootless mode failed. Others passed.
lnie fcos next image,x86_64,VM created by testcloud
Pass pass
Pass pass
Pass pass
Pass pass
lnie fcos next image,x86_64,VM created with virt-install
Pass pass
lruzicka QEMU
Pass pass
Pass pass
Pass pass
Pass pass
Pass pass
Pass pass
[1]
Pass pass
Pass pass
Pass pass
Pass pass
Pass pass
  1. The vim package cannot be installed because there is a conflict of versions, as far as I can tell from the output in journalctl when attempting to install it. However, the mechanism does work normally when trying to install a different editor, such as e3.
sohank2602
Pass pass
[1]
  1. After discussing with Dusty on IRC, followed the first three steps from https://fedoraproject.org/wiki/QA:Testcase_CoreOS_swap_on_zram to verify this test case

Upgrade[edit]

User Profile Switch stream Bootloader updates References
alciregi DigitalOcean droplet
Pass pass
Pass pass
donaldsebleung FCOS stable image with QEMU; arch: x86_64; custom Ignition file with password-based SSH access; launching with QEMU
Warning warn
[1]
Pass pass
[2]
  1. FCOS stable prints IP address of VM to console but FCOS next does not after a switch stream. No other issues found with FCOS next
  2. Switch stream stable -> next successful. Did not test rollback.
donaldsebleung FCOS stable image with QEMU; arch: x86_64; custom Ignition file with password-based SSH access; launching with QEMU
Pass pass
[1]
  1. Bootloader updates behaved as expected. Did not test automatic bootloader updates with systemd service
dustymabe FCOS stable image with qemu - rebased to next.
Pass pass
Pass pass
hricky Intel® Core™ i7-4790 CPU, 32 GB RAM, Bare Metal
Pass pass
[1]
Pass pass
  1. next → testing → stable
lruzicka QEMU
Pass pass
Pass pass

Tutorials[edit]

User Profile References

Miscellaneous[edit]

User Profile Documentation Exploratory testing References
Nemric PXE installation
Pass pass
[1]
Pass pass
[2]
  1. Setting up journald to run in RAM (since promtail send logs to grafana/loki) with : `yaml systemd: units: - name: systemd-journald.service dropins: - name: 11-storage_volatile.conf contents: | [Journal] Storage=volatile RuntimeMaxUse=16M RuntimeMaxFileSize=16M RuntimeMaxFiles=1 - name: systemd-journal-flush.service mask: true`
  2. Trying the rewritten fscahe that landed on kernel 5.17 -> sudo rpm-ostree install --apply-live cachefilesd && sudo systemctl enable --now cachefilesd Seems to be more reliable : it worked at first attempt and still alive (cat /proc/fs/nfsfs/* : fsc = yes) after nfs server reboot and/or host reboot
hricky Intel® Core™ i7-4790 CPU, 32 GB RAM, Bare Metal
Pass pass
lruzicka QEMU
Fail fail
[1]
  1. The approach in this procedure does not work https://docs.fedoraproject.org/en-US/fedora-coreos/provisioning-libvirt/. Any attempt fails with "ERROR internal error: qemu ended monitor unexpectedly: 2022-04-11T11:17:53.116624Z qemu-system-x86_64: -fw_cfg name=opt/com.coreos/config,file=/home/lruzicka/todelete/coreos_yaml.ign: can't load /home/lruzicka/todelete/coreos.ign: Failed to open file “/home/lruzicka/todelete/coreos_yaml.ign”: Permission denied. "