- log_href:
http://qa-proxy.ceph.com/teuthology/lflores-2024-06-10_22:21:08-rados-wip-lflores-testing-2-2024-06-10-1925-distro-default-smithi/7749762/teuthology.log
- archive_path:
/home/teuthworker/archive/lflores-2024-06-10_22:21:08-rados-wip-lflores-testing-2-2024-06-10-1925-distro-default-smithi/7749762
- description:
rados/cephadm/workunits/{0-distro/ubuntu_22.04 agent/off mon_election/classic task/test_host_drain}
- duration:
0:23:30
- email:
- failure_reason:
- flavor:
default
- job_id:
7749762
- kernel:
- last_in_suite:
False
- machine_type:
smithi
- name:
lflores-2024-06-10_22:21:08-rados-wip-lflores-testing-2-2024-06-10-1925-distro-default-smithi
- nuke_on_error:
- os_type:
ubuntu
- os_version:
22.04
- overrides:
- admin_socket:
- branch:
wip-lflores-testing-2-2024-06-10-1925
- ceph:
- conf:
- global:
- mon election default strategy:
1
- mgr:
- debug mgr:
20
- debug ms:
1
- mgr/cephadm/use_agent:
False
- mon:
- debug mon:
20
- debug ms:
1
- debug paxos:
20
- osd:
- debug ms:
1
- debug osd:
20
- flavor:
default
- log-ignorelist:
-
\(MDS_ALL_DOWN\)
-
\(MDS_UP_LESS_THAN_MAX\)
-
MON_DOWN
-
mons down
-
mon down
-
out of quorum
-
CEPHADM_STRAY_HOST
-
CEPHADM_STRAY_DAEMON
- log-only-match:
- sha1:
37bbe1d8d788b7c9a71280e7d510c295a5853f0b
- ceph-deploy:
- conf:
- client:
- log file:
/var/log/ceph/ceph-$name.$pid.log
- mon:
- install:
- ceph:
- flavor:
default
- sha1:
37bbe1d8d788b7c9a71280e7d510c295a5853f0b
- workunit:
- branch:
wip-lflores-testing-2-2024-06-10-1925
- sha1:
37bbe1d8d788b7c9a71280e7d510c295a5853f0b
- owner:
scheduled_lflores@teuthology
- pid:
- roles:
-
['host.a', 'mon.a', 'mgr.a', 'osd.0', 'osd.1']
-
['host.b', 'mon.b', 'mgr.b', 'osd.2', 'osd.3']
-
['host.c', 'mon.c', 'osd.4', 'osd.5']
- sentry_event:
- status:
pass
- success:
True
- branch:
wip-lflores-testing-2-2024-06-10-1925
- seed:
7590
- sha1:
37bbe1d8d788b7c9a71280e7d510c295a5853f0b
- subset:
111/120000
- suite:
rados
- suite_branch:
wip-lflores-testing-2-2024-06-10-1925
- suite_path:
/home/teuthworker/src/git.ceph.com_ceph-c_37bbe1d8d788b7c9a71280e7d510c295a5853f0b/qa
- suite_relpath:
qa
- suite_repo:
https://git.ceph.com/ceph-ci.git
- suite_sha1:
37bbe1d8d788b7c9a71280e7d510c295a5853f0b
- targets:
- smithi043.front.sepia.ceph.com:
ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBIr2EytLqu8MGRTGWagGAW0p5jlYeMqvk06/FQjb8/wTaIx9QvVT2X78dsTlF83ZUSMliwrRaiEr/qK0ZwiVPZQ=
- smithi094.front.sepia.ceph.com:
ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBBo4Jx/gRPc7NHjg2pDMP7TBubwNB/2jCfyJ67hrfZWqb4Blu1GSYvaghNeypakuhKDIxEx/+8vM4eHeu35OZJU=
- smithi133.front.sepia.ceph.com:
ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBOnCEU07AxSKR+siq1DZlSI/RsxoRX1TJZSMeLPCPeW/VzyKWPDi3wYfi8KwUC8D/RD7bBn1dtAH2i+6LqZUWl0=
- tasks:
-
- internal.buildpackages_prep:
-
- 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:
-
- internal.base:
-
- internal.archive_upload:
-
- internal.archive:
-
- internal.coredump:
-
- internal.sudo:
-
- internal.syslog:
-
- internal.timer:
-
- pcp:
-
- selinux:
-
- ansible.cephlab:
-
- clock:
-
- install:
-
- cephadm:
-
- cephadm.shell:
- host.a:
-
set -ex
HOSTNAMES=$(ceph orch host ls --format json | jq -r '.[] | .hostname')
for host in $HOSTNAMES; do
# find the hostname for "host.c" which will have no mgr
HAS_MGRS=$(ceph orch ps --hostname ${host} --format json | jq 'any(.daemon_type == "mgr")')
if [ "$HAS_MGRS" == "false" ]; then
HOST_C="${host}"
fi
done
# One last thing to worry about before draining the host
# is that the teuthology test tends to put the explicit
# hostnames in the placement for the mon service.
# We want to make sure we can drain without providing
# --force and there is a check for the host being removed
# being listed explicitly in the placements. Therefore,
# we should remove it from the mon placement.
ceph orch ls mon --export > mon.yaml
sed /"$HOST_C"/d mon.yaml > mon_adjusted.yaml
ceph orch apply -i mon_adjusted.yaml
# now drain that host
ceph orch host drain $HOST_C --zap-osd-devices
# wait for drain to complete
HOST_C_DAEMONS=$(ceph orch ps --hostname $HOST_C)
while [ "$HOST_C_DAEMONS" != "No daemons reported" ]; do
sleep 15
HOST_C_DAEMONS=$(ceph orch ps --hostname $HOST_C)
done
# we want to check the ability to remove the host from
# the CRUSH map, so we should first verify the host is in
# the CRUSH map.
ceph osd getcrushmap -o compiled-crushmap
crushtool -d compiled-crushmap -o crushmap.txt
CRUSH_MAP=$(cat crushmap.txt)
if ! grep -q "$HOST_C" <<< "$CRUSH_MAP"; then
printf "Expected to see $HOST_C in CRUSH map. Saw:\n\n$CRUSH_MAP"
exit 1
fi
# If the drain was successful, we should be able to remove the
# host without force with no issues. If there are still daemons
# we will get a response telling us to drain the host and a
# non-zero return code
ceph orch host rm $HOST_C --rm-crush-entry
# verify we've successfully removed the host from the CRUSH map
sleep 30
ceph osd getcrushmap -o compiled-crushmap
crushtool -d compiled-crushmap -o crushmap.txt
CRUSH_MAP=$(cat crushmap.txt)
if grep -q "$HOST_C" <<< "$CRUSH_MAP"; then
printf "Saw $HOST_C in CRUSH map after it should have been removed.\n\n$CRUSH_MAP"
exit 1
fi
teuthology_branch:
main
verbose:
True
pcp_grafana_url:
priority:
100
user:
lflores
queue:
posted:
2024-06-10 22:23:54
started:
2024-06-11 17:29:49
updated:
2024-06-11 18:04:56
status_class:
success
runtime:
0:35:07
wait_time:
0:11:37