Description: upgrade:hammer-x/v0-94-4-stop/{distros/centos_latest.yaml distros/ubuntu_14.04.yaml ignore.yaml v0-94-4-stop.yaml}

Log: http://qa-proxy.ceph.com/teuthology/yuriw-2018-05-22_17:03:59-upgrade:hammer-x-jewel-distro-basic-vps/2567761/teuthology.log

  • log_href: http://qa-proxy.ceph.com/teuthology/yuriw-2018-05-22_17:03:59-upgrade:hammer-x-jewel-distro-basic-vps/2567761/teuthology.log
  • archive_path: /home/teuthworker/archive/yuriw-2018-05-22_17:03:59-upgrade:hammer-x-jewel-distro-basic-vps/2567761
  • description: upgrade:hammer-x/v0-94-4-stop/{distros/centos_latest.yaml distros/ubuntu_14.04.yaml ignore.yaml v0-94-4-stop.yaml}
  • duration:
  • email: ceph-qa@ceph.com
  • failure_reason:
  • flavor:
  • job_id: 2567761
  • kernel:
    • sha1: distro
    • kdb: True
  • last_in_suite: False
  • machine_type: vps
  • name: yuriw-2018-05-22_17:03:59-upgrade:hammer-x-jewel-distro-basic-vps
  • nuke_on_error: True
  • os_type: ubuntu
  • os_version: 14.04
  • overrides:
    • ceph:
      • log-whitelist:
        • \(MDS_ALL_DOWN\)
        • \(MDS_UP_LESS_THAN_MAX\)
        • scrub mismatch
        • ScrubResult
      • conf:
        • mon:
          • mon warn on legacy crush tunables: False
          • debug mon: 20
          • debug paxos: 20
          • debug ms: 1
        • osd:
          • debug osd: 25
          • debug filestore: 20
          • debug journal: 20
          • debug ms: 1
      • sha1: 98aa3b683021b0d7329319bbc74736d777603968
    • ceph-deploy:
      • conf:
        • client:
          • log file: /var/log/ceph/ceph-$name.$pid.log
        • mon:
          • osd default pool size: 2
    • workunit:
      • sha1: 98aa3b683021b0d7329319bbc74736d777603968
      • branch: jewel
    • install:
      • ceph:
        • sha1: 98aa3b683021b0d7329319bbc74736d777603968
    • admin_socket:
      • branch: jewel
  • owner: scheduled_yuriw@teuthology
  • pid:
  • roles:
    • ['mon.a', 'osd.0', 'osd.1']
    • ['osd.2']
  • sentry_event:
  • status: dead
  • success:
  • branch: jewel
  • seed:
  • sha1: 98aa3b683021b0d7329319bbc74736d777603968
  • subset:
  • suite:
  • suite_branch: jewel
  • suite_path:
  • suite_relpath:
  • suite_repo:
  • suite_sha1: 98aa3b683021b0d7329319bbc74736d777603968
  • targets:
  • tasks:
    • internal.check_packages:
    • internal.buildpackages_prep:
    • internal.lock_machines:
      • 2
      • vps
    • internal.save_config:
    • internal.check_lock:
    • internal.add_remotes:
    • console_log:
    • internal.connect:
    • internal.push_inventory:
    • internal.serialize_remote_roles:
    • internal.check_conflict:
    • internal.check_ceph_data:
    • internal.vm_setup:
    • kernel:
      • sha1: distro
      • kdb: True
    • internal.base:
    • internal.archive_upload:
    • internal.archive:
    • internal.coredump:
    • internal.sudo:
    • internal.syslog:
    • internal.timer:
    • pcp:
    • selinux:
    • ansible.cephlab:
    • clock:
    • print: **** Install version lower than v0.94.4
    • install:
      • tag: v0.94.3
    • ceph:
      • fs: xfs
    • print: *** Upgrade the target that runs osd.0 and osd.1 to -x while the target that runs osd.2 stays v0.94.3
    • install.upgrade:
      • osd.0:
    • print: *** Restart the mon.a so that it is post-hammer v0.94.4 and implements the CEPH_FEATURE_HAMMER_0_94_4 feature
    • ceph.restart:
      • daemons:
        • mon.a
    • print: *** Verify that osd.0 cannot restart because osd.1 and osd.2 are still < v0.94.4
    • ceph.restart:
      • daemons:
        • osd.0
      • wait-for-healthy: False
    • exec:
      • osd.0:
        • set -x success=false for delay in 1 2 4 8 16 32 64 128 256 512 1024 ; do if ceph daemon osd.0 log flush ; then if grep "one or more pre-v0.94.4 hammer" /var/log/ceph/ceph-osd.0.log ; then success=true break fi fi sleep $delay done $success || exit 1
    • print: *** Stop all OSDs and restart osd.0 and osd.1 which are >= v0.94.4
    • ceph.stop:
      • daemons:
        • osd.0
        • osd.1
        • osd.2
    • exec:
      • mon.a:
        • set -x ceph osd down osd.0 ceph osd down osd.1 ceph osd down osd.2
    • ceph.restart:
      • daemons:
        • osd.0
        • osd.1
      • wait-for-healthy: False
    • exec:
      • mon.a:
        • set -x success=false for delay in 1 2 4 8 16 32 64 128 256 512 1024 ; do if ceph osd dump | grep 'osd.1 up' && ceph osd dump | grep 'osd.0 up' ; then success=true break fi ceph osd dump sleep $delay done $success || exit 1 ceph osd dump | grep 'osd.2 down' || exit 1
    • print: *** Verify that osd.2 cannot restart because it is < v0.94.4 and all other OSDs are >= v0.94.4
    • ceph.restart:
      • daemons:
        • osd.2
      • wait-for-healthy: False
    • exec:
      • mon.a:
        • set -x success=false for delay in 1 2 4 8 16 32 64 128 256 512 1024 ; do ceph daemon mon.a log flush if grep "disallowing boot of pre-hammer v0.94.4 OSD" /var/log/ceph/*.log ; then success=true break fi sleep $delay ceph osd dump done $success || exit 1
    • print: *** Upgrade the target that runs osd.2 to -x and verify the cluster is back to being healthy
    • install.upgrade:
      • osd.2:
    • ceph.restart:
      • daemons:
        • osd.2
      • wait-for-healthy: False
    • exec:
      • mon.a:
        • sleep 300
        • ceph osd set sortbitwise
        • ceph osd set require_jewel_osds
    • ceph.healthy:
  • teuthology_branch: master
  • verbose: True
  • pcp_grafana_url:
  • priority:
  • user:
  • queue:
  • posted: 2018-05-22 17:04:12
  • started: 2018-05-22 17:07:15
  • updated: 2018-05-22 23:15:20
  • status_class: danger
  • runtime: 6:08:05