QA (public) » openQA Project (public) » openQA Infrastructure (public)

I added the salt key for worker13.oqa.suse.de one of the recently migrated workers now. And now sudo salt --no-color --state-output=changes '*' grains.get fqdn yields

storage.qa.suse.de:
    storage
openqaworker2.suse.de:
    openqaworker2.suse.de
openqaworker9.suse.de:
    openqaworker9.suse.de
openqaworker8.suse.de:
    openqaworker8.suse.de
openqaworker3.suse.de:
    openqaworker3.suse.de
openqaworker14.qa.suse.cz:
    openqaworker14.qa.suse.cz
qamasternue.qa.suse.de:
    qamasternue.qa.suse.de
QA-Power8-5-kvm.qa.suse.de:
    QA-Power8-5-kvm.qa.suse.de
baremetal-support:
    baremetal-support.qa.suse.de
tumblesle:
    tumblesle.qa.suse.de
openqaworker6.suse.de:
    openqaworker6.suse.de
jenkins.qa.suse.de:
    jenkins
openqaworker5.suse.de:
    openqaworker5.suse.de
openqa.suse.de:
    openqa.suse.de
schort-server:
    schort-server.qa.suse.de
openqa-monitor.qa.suse.de:
    monitor.qa.suse.de
malbec.arch.suse.de:
    malbec.arch.suse.de
backup.qa.suse.de:
    backup-vm
worker13.oqa.suse.de:
    worker13
QA-Power8-4-kvm.qa.suse.de:
    QA-Power8-4-kvm.qa.suse.de
grenache-1.qa.suse.de:
    grenache-1.qa.suse.de
openqaworker-arm-1.suse.de:
    openqaworker-arm-1.suse.de
openqaworker-arm-2.suse.de:
    openqaworker-arm-2.suse.de
openqaworker-arm-3.suse.de:
    openqaworker-arm-3.suse.de

so the FQDN of worker13 and also others is not really "full", yet. https://stackoverflow.com/a/54750233 recommends to avoid FQDN and in nearly all cases our salt ID is already equivalent to the FQDN so this might be a better choice. sudo salt --no-color --state-output=changes -C 'G@roles:worker' grains.get fqdn returns a good match for all except openqaworker2 which is currently not with a working DNS entry and worker13 where hostname -f does not return a match. Also sudo salt --no-color --state-output=changes -C 'G@roles:worker' cmd.run 'nslookup $(hostname)' shows that workers can fully resolve themselves except for worker13, the mentioned problematic one.

hosts within the new domain .oqa.suse.de. should search for matches within that domain so that nslookup $(hostname) works, e.g. nslookup worker13 should work. I assume that salt is relying on that to return a proper match for grains.fqdn

https://gitlab.suse.de/openqa/salt-states-openqa/-/merge_requests/766

https://gitlab.suse.de/openqa/salt-states-openqa/-/merge_requests/766 merged. https://gitlab.suse.de/openqa/salt-pillars-openqa/-/merge_requests/451 for related cleanup in pillars and also using FQDN for SUT_IP and VIRSH_GUEST.

There are some problems on s390x kvm tests, e.g. can't find login prompt in https://openqa.suse.de/tests/9900527 or unable to connect to VNC like in https://openqa.suse.de/tests/9900606#step/installation/26 . I removed worker2 from salt and did:

for i in $(sed -n 's/VIRSH_GUEST=//p' /etc/openqa/workers.ini | sort | uniq | grep 's390kvm'); do echo $i && sed -i "s@$i@$(host $i | sed -n 's/^.*has address //p')@" /etc/openqa/workers.ini; done

and retriggered some tests, let's see, e.g. https://openqa.suse.de/tests/9901355

EDIT: Tests passed the installation and continue fine so the hostname is apparently a problem.

I changed back only on instance 32 on worker2 the SUT_IP to hostname, keeping VIRSH_GUEST on the IPv4 address and added a special worker class
openqa-clone-job --within-instance https://openqa.suse.de/tests/9901355 _GROUP=0 BUILD= TEST=qam-gnome-test-okurz-poo109241 WORKER_CLASS=debug_okurz_poo109241
Created job #9901761: sle-12-SP4-Server-DVD-Updates-s390x-Build20221107-1-qam-gnome@s390x-kvm-sle12 -> https://openqa.suse.de/t9901761

https://gitlab.suse.de/openqa/salt-pillars-openqa/-/merge_requests/455

@okurz Would you help have a look at this issue ? Thanks
https://gitlab.suse.de/openqa/salt-pillars-openqa/-/merge_requests/481