e29051fd9c
Still seeing oom-killer trigger in Cinder Ceph CI jobs with ceph-osd being a top memory consumer. It appears to default to trying to use 4GB of memory, change this to 2GB to see if it helps. Increases the job timeout to 3h for a bit more headroom since jobs are regularly taking more than 2h to complete. Change-Id: I71a46a452914256f36299623a29ed7eebf8e61d2
40 lines
1.5 KiB
Bash
40 lines
1.5 KiB
Bash
# cephadm.sh - DevStack extras script to install Ceph
|
|
if [[ "$1" == "source" ]]; then
|
|
# Initial source
|
|
source $TOP_DIR/lib/cephadm
|
|
elif [[ "$1" == "stack" && "$2" == "pre-install" ]]; then
|
|
if [[ "$ENABLE_CEPH_RGW" = "True" ]] && (is_service_enabled swift); then
|
|
die $LINENO \
|
|
"You cannot activate both Swift and Ceph Rados Gateway, \
|
|
please disable Swift or set ENABLE_CEPH_RGW=False"
|
|
fi
|
|
echo_summary "[cephadm] Configuring system services ceph"
|
|
pre_install_ceph
|
|
elif [[ "$1" == "stack" && "$2" == "install" ]]; then
|
|
if [[ "$REMOTE_CEPH" = "False" ]]; then
|
|
# Perform installation of service source
|
|
echo_summary "[cephadm] Installing ceph"
|
|
install_ceph
|
|
set_memory_config
|
|
set_min_client_version
|
|
else
|
|
echo "[CEPHADM] Remote Ceph: Skipping install"
|
|
get_cephadm
|
|
fi
|
|
elif [[ "$1" == "stack" && "$2" == "post-config" ]]; then
|
|
echo_summary "[cephadm] Configuring additional Ceph services"
|
|
configure_ceph
|
|
if [[ "$MDS_LOGS" == "True" ]]; then
|
|
enable_verbose_mds_logging
|
|
fi
|
|
elif [[ "$1" == "stack" && "$2" == "test-config" ]]; then
|
|
if is_service_enabled tempest; then
|
|
iniset $TEMPEST_CONFIG compute-feature-enabled swap_volume False
|
|
iniset $TEMPEST_CONFIG compute-feature-enabled shelve True
|
|
iniset $TEMPEST_CONFIG volume-feature-enabled extend_attached_volume True
|
|
iniset $TEMPEST_CONFIG volume-feature-enabled volume_revert True
|
|
fi
|
|
elif [[ "$1" == "unstack" || "$1" == "clean" ]]; then
|
|
cleanup_ceph
|
|
fi
|