QA » openQA Project » openQA Tests » ALP

Transactional Update test description¶

Basic checks¶

microos/networking.pm¶

Basic network and connectivity check.

• Print network configuration (ip addr show and resolv.conf)
• Ping localhost 127.0.0.1 (IPv4) and ::1 (IPv6)
• Curl to openqa.opensuse.org and github.com

microos/libzypp_config.pm¶

Check that zypper configuration is customized for MicroOS.

• egrep -x "^solver.onlyRequires ?= ?true" /etc/zypp/zypp.conf
• egrep -x "^rpm.install.excludedocs ?= ?yes" /etc/zypp/zypp.conf
• egrep -x "^multiversion ?=" /etc/zypp/zypp.conf

microos/image_checks.pm¶

Simple image specific checks:

• Verify that openQA resized the disk image
• Verify that there is no unpartitioned space left
• Verify that the filesystem mounted at /var grew beyond the default 5GiB
• Verify that combustion ran

microos/one_line_checks.pm¶

Simple checks after installation.

• Check that snapper is configured snapper list
• Subvolume check btrfs subvolume show /var
• kubectl is not installable if SYSTEM_ROLE=microos (only openSUSE MicroOS)
• crio should be installed if SYSTEM_ROLE=kubeadm (only openSUSE MicroOS)

microos/services_enabled.pm¶

Check that services are enables based on system role
The following services should be enabled by default:

• issue-add-ssh-keys
• issue-generator
• transactional-update.timer
• sshd

On other system roles (MicroOS only), other services might be checked to be enabled.

microos/cockpit_service.pm¶

Basic check for cockpit service.

• Install cockpit package if not installed
• Install cockpit-networkmanager package if the system is running network-manager.
• Install cockpit-wicked package if the system is running wicked.
• Install cockpit-machines and cockpit-tukit if packages are available.
• Enable and start cockpit.socket service and check that Cockpit is active and accessible on http://localhost:9090
• Test Cockpit service survives a reboot (it's active automatically after rebooting)

Specific transactional system tests¶

transactional/trup_smoke.pm¶

Basic smoke test to verify all basic transactional update operations work and system can properly boot.

• Reinstall bootloader: transactional-update bootloader
• Regenerate grub.cfg: transactional-update grub.cfg
• Regenerate initrd: transactional-update initrd
• Regenerate kdump: transactional-update kdump
• Run cleanup: transactional-update cleanup

transactional/filesystem_ro.pm¶

Check that root filesystem is read only

• / is RO:
  • cannot create touch /should_fail
  • btrfs property get / ro | grep "ro=true"
• /var is RW:
  • can create touch /var/log/should_succeed
  • btrfs property get /var ro | grep "ro=false"
• /etc is RW:
  • can create touch /etc/should_succeed
• Look for ro mount point in fstab: findmnt -s / -n -O ro
• Look for ro mount point in mounted filesystems: findmnt / -n -O ro

transactional/transactional_update.pm¶

TBD

transactional/rebootmgr.pm¶

Test rebootmgr using different strategies

• Test instant reboot
• Test maint-window strategy
• Test etcd locking strategy

transactional/health_check.pm¶

Check that health-check service works correctly.

• Install and enable service health-checker if not installed.
• health-checker command shall return passed string.
• Current snapshot shall be the same as the logged one from health-checker.
• Update rebootmgr.sh using transactional update shell to force health-checker to fail:
  • f=$(rpm --eval %{_libexecdir})/health-checker/fail.sh; echo -e \'#/bin/sh\n[ "$1" != "check" ]\' > $f && chmod a+x $f
  • reboot
  • health-checker should rollback to the correct snapshot
• health-checker command shall return passed string.