action #19122
Updated by asmorodskyi about 7 years ago
## Setup story Since we often need a architecture or machine specific change to the test code or change in test code that is likely to behave different depending on the architecture it makes sense to have development instances for non-hammer architecture available as well besides the production instances. I am trying to setup a ppc64le worker on shiraz-2, a virtual machine running on shiraz.arch.suse.de. Steps I did so far to progress: * Install SLES 12 SP2 * Install qemu-system-ppc64le from standard SLES repo * Install openQA from devel:openQA * ``` zypper ar -r http://download.opensuse.org/repositories/Virtualization/SLE_12/Virtualization.repo modprobe kvm kvm_pr zypper in qemu-vgabios qemu-ipxe rsync -aHP shiraz-3:/usr/share/qemu/slof.bin /usr/share/qemu/slof.bin ``` /usr/share/qemu/slof.bin has md5sum dee802a9f7d906c49150c6e6f61240a0 that failed now in http://lord.arch/tests/6203#step/boot_to_desktop_sym/7 with "Facility 'TM' unavailable, exception" and kernel panic. Trying with qemu from Virtualization repo which also overwrote /usr/share/qemu/slof.bin (md5sum 5cc28a978f5b6ff657c9ed2779ceb3e3). Seems to work better now http://lord.arch/tests/6204#live ## Worker configuration * login as user to shiraz-2 * Use sudo (without password) to configure the openQA worker configuration files * add your host in workers.ini * add credentials in client.conf * Make sure your webUI host delivers tests+assets over e.g. rsync server (without authentication read-only) * Restart worker * Ready to accept jobs Example setup: ``` vim /etc/openqa/workers.ini /etc/openqa/client.conf ``` Test rsync with /etc/rsync.conf example ``` gid = nobody uid = nobody read only = true use chroot = true transfer logging = true log format = %h %o %f %l %b log file = /var/log/rsyncd.log pid file = /var/run/rsyncd.pid slp refresh = 300 use slp = false [tests] path = /var/lib/openqa/share/tests comment = OpenQA Test Distributions ``` rsync rsync://<your_webui_host>/tests ``` it should list tests directory content without asking for authentication. Then restart workers 1 and 2, make sure they don't show errors and clone a job.