29 Jan 2025
Fedora People
Fedora Infrastructure Status: Updates and Reboots
29 Jan 2025 9:00pm GMT
24 Jan 2025
Fedora People
Fedora Community Blog: Infra and RelEng Update – Week 04 2025
This is a weekly report from the I&R (Infrastructure & Release Engineering) Team. We provide you both infographic and text version of the weekly report. If you just want to quickly look at what we did, just look at the infographic. If you are interested in more in depth details look below the infographic.
Week: 20th January - 24th January 2025
Infrastructure & Release Engineering
The purpose of this team is to take care of day to day business regarding CentOS and Fedora Infrastructure and Fedora release engineering work.
It's responsible for services running in Fedora and CentOS infrastructure and preparing things for the new Fedora release (mirrors, mass branching, new namespaces etc.).
List of planned/in-progress issues
Fedora Infra
- In progress:
- Please generate the Fedora 44 gpg key
- Opt-in for AWS region to mx-central-1 acct:125523088429
- Opt-in for AWS region to ap-southeast-7 acct:125523088429
- retire easyfix
- bvmhost-p09-03 ends in emergency mode
- Add StartInstanceRefresh permission for fedora-ci-testing-farm user in AWS
- distgit-bugzilla-sync poddlers in both staging and prod downloading a lot from production
- Deploy Element Server Suite operator in staging
- Create CentOS calendar
- Pagure returns error 500 trying to open a PR on https://src.fedoraproject.org/rpms/python-setuptools-gettext
- Inactive packagers policy for the F41 release cycle
- a real domain name for konflux
- Retirement of `monitor-dashboard` tracker
- Create a POC integration in Konflux for fedora-infra/webhook-to-fedora-messaging
- Retirement of `monitor-gating` tracker
- setup ipa02.stg and ipa03.stg again as replicas
- Manage our new testing.farm domain via AWS Route53
- Add support for creating RDS instances under `testing-farm-` prefix
- Move OpenShift apps from deploymentconfig to deployment
- RFE: fedoras container image register change
- The process to update the OpenH264 repos is broken
- httpd 2.4.61 causing issue in fedora infrastructure
- Support allocation dedicated hosts for Testing Farm
- fedorapeople.org directory listing theme needs refreshing
- EPEL minor version archive repos in MirrorManager
- vmhost-x86-copr01.rdu-cc.fedoraproject.org DOWN
- Add yselkowitz to list to notify when ELN builds fail
- Cleaning script for communishift
- rhel7 eol
- move resultsdb-ci-listener deployment pipeline
- rhel7 EOL - github2fedmsg
- Move from iptables to firewalld
- Help me move my discourse bots to production?
- fedmsg -> fedora-messaging migration tracker
- rhel9 adoption
- Create monitoring tool for rabbitmq certificates
- Replace Nagios with Zabbix in Fedora Infrastructure
- Migration of registry.fedoraproject.org to quay.io
- Commits don't end up on the scm-commits list
- Done:
- Move personal GitHub repository to fedora-infra org
- extra trailing slash on epel.io redirect breaks page rendering
- Messed up rpm/valgrind pull request/branch
- New Issue information page broken link
- bastion delivering locally to non contributor accounts
- mirrorlist is returning too few mirrors
- New requirements for Google and Yahoo mail at least
CentOS Infra including CentOS CI
- In progress:
- Done:
Release Engineering
- In progress:
- Unretire tachyon
- Unretire ethos
- Quay.io repository for fedora-bootc-tier-x
- Missing rawhide branch for recently created package 'pybind11-json'
- F42 Self-Contained Change: Switch to EROFS for Live Media
- Please remove two branches from octave-iso2mesh
- Send compose reports to a to-be-created separate ML
- .sqlite metadata missing in f41-updates and f41-updates-testing repositories
- F42 system-wide change: GNU Toolchain update for F42 https://fedoraproject.org/wiki/Changes/GNUToolchainF42
- Delete "firefox-fix" branch in xdg-desktop-portal
- please create epel10 based el10-openjdk tag
- Could we have fedoraproject-updates-archive.fedoraproject.org for Rawhide?
- Investigate and untag packages that failed gating but were merged in via mass rebuild
- a few mass rebuild bumps failed to git push - script should retry or error
- Renaming distribution media for Fedora Server
- Package retirements are broken in rawhide
- Implement checks on package retirements
- Untag containers-common-0.57.1-6.fc40
- orphan-all-packages.py should remove bugzilla_contact entries from fedora-scm-requests as well
- Packages that fail to build SRPM are not reported during the mass rebuild bugzillas
- When orphaning packages, keep the original owner as co-maintainer
- Create an ansible playbook to do the mass-branching
- RFE: Integration of Anitya to Packager Workflow
- Cleaning old stuff from koji composes directories
- Fix tokens for ftbfs_weekly_reminder. script
- Update bootloader components assignee to "Bootloader Engineering Team"for Improved collaboration
- Done:
- Stalled EPEL package: python-zope-exceptions
- Stalled EPEL package: python-pytest-subtests
- F42 Self-Contained Change: Promote KDE Plasma Desktop variant to Edition
- The build `localsearch-3.8~rc-1.fc42` is in `DELETED` state and is blocking me from rebuilding
- Failed request-repo action for sgx-rpm-macros
- Unretire virt-who
- latest-Fedora-Cloud-41 directory missing on the cloud compose listing
If you have any questions or feedback, please respond to this report or contact us on #redhat-cpe channel on matrix.
The post Infra and RelEng Update - Week 04 2025 appeared first on Fedora Community Blog.
24 Jan 2025 10:00am GMT
Fedora Magazine: Update on hibernation in Fedora Workstation
Goals and rationale
Hibernation stores the state of the whole operating system - the contents of memory used by the kernel and all programs - on disk. The machine is then completely powered off. Upon next boot, this state is restored and the old kernel and all the programs that were running continue execution.
Hibernation is nowadays used less often, because "suspend" - the state where CPU is powered down, but the contents of memory are preserved, works fine on most laptops and other small devices. But if the suspend is implemented poorly and it drains the battery too quickly, or if the user needs to completely power off the device for some reasons, hibernation can still be useful.
We need a storage area for hibernation. The kernel allows two options:
- either a single large-enough swap device, usually a partition,
- or a single large-enough swap file on some file system.
Fedora Linux installations by default do not use a normal swap device or file. Instead, a zram device is created, which is an in-memory compressed swap area. It is not suitable for hibernation. This means that hibernation does not work out-of-the-box on Fedora Linux. This guide describes how to create a swap file to enable hibernation.
Limitations
This method only works on UEFI!
To check that the system uses UEFI:
bootctl
If this commands prints "Not booted with EFI", then the method described below won't work. Refer to the original Hibernation in Fedora Workstation (for Fedora Linux 36) instead.
Another severe limitation is that SecureBoot must be disabled. The kernel does not allow hibernation with SecureBoot! Disabling SecureBoot reduces the security of the machine somewhat. Thus, do this only if you think that hibernation is more worth it.
Implementation
First, check whether Secure Boot is on:
bootctl
If this prints "Secure Boot: disabled" then SB is off. Otherwise, reboot the machine, go into UEFI settings, and disable Secure Boot.
Second, create and enable a swap file:
SWAPSIZE=$(free | awk '/Mem/ {x=$2/1024/1024; printf "%.0fG", (x<2 ? 2*x : x<8 ? 1.5*x : x) }') sudo btrfs subvolume create /var/swap sudo mkswap --file -L SWAPFILE --size $SWAPSIZE /var/swap/swapfile sudo bash -c 'echo /var/swap/swapfile none swap defaults 0 0 >>/etc/fstab' sudo swapon -av
This should print a message that swap was enabled on /var/swap/swapfile. The swap file is added to fstab, so it'll be permanently active. This is a good thing, it should make the system more reliable in general.
Now we are ready to test hibernation:
systemctl hibernate
After the system has shut down, boot it again and let one of the kernels start. The machine should return to the previous state from before hibernation.
This method does not require further configuration because systemd automatically stores the location of the swap file before entering hibernation in an UEFI variable, and then after the reboot, reads that variable and instruct the kernel to resume from this location. This only works on UEFI systems, but is otherwise quite simple and robust.
Reverting the changes
sudo swapoff -v /var/swap/swapfile sudo sed -r -i '/.var.swap.swapfile/d' /etc/fstab sudo btrfs subvolume rm /var/swap
After that, reenable SecureBoot if appropriate.
Troubleshooting
This process mail fail in two ways:
- either going into hibernation fails, i.e. the kernel does not save the state and the machine does not actually power off,
- or loading of saved state fails, and we end up with a fresh boot.
In both cases, the first step is to look at journalctl -b, in particular any error lines.
24 Jan 2025 8:00am GMT