Restart all OSDs and do not stop after the first one.

The current handler only restarts one OSD on each OSD server. After
the first one the handler stops, not matter what results the checks had.

Co-Authored-By: Gaudenz Steinlin (@gaudenz)

diff --git a/roles/ceph-common/templates/restart_osd_daemon.sh.j2 b/roles/ceph-common/templates/restart_osd_daemon.sh.j2
index 5d1d04597b52a9de9eeae84855f5ea949c314ed4..f027574cf6d2c7244203f0a9a81c01f37e727d1d 100644 (file)
--- a/roles/ceph-common/templates/restart_osd_daemon.sh.j2
+++ b/roles/ceph-common/templates/restart_osd_daemon.sh.j2
@@ -8,7 +8,7 @@ check_pgs() {
   while [ $RETRIES -ne 0 ]; do
     test "[""$(ceph $CEPH_CLI -s -f json | python -c 'import sys, json; print(json.load(sys.stdin)["pgmap"]["num_pgs"])')""]" == "$(ceph $CEPH_CLI -s -f json | python -c 'import sys, json; print [ i["count"] for i in json.load(sys.stdin)["pgmap"]["pgs_by_state"] if i["state_name"] == "active+clean"]')"
     RET=$?
-    test $RET -eq 0 && exit 0
+    test $RET -eq 0 && return 0
     sleep $DELAY
     let RETRIES=RETRIES-1
   done
@@ -28,7 +28,7 @@ for id in $(ls /var/lib/ceph/osd/ | sed 's/.*-//'); do
   # Wait and ensure the socket exists after restarting the daemon
   SOCKET=/var/run/ceph/{{ cluster }}-osd.${id}.asok
   while [ $COUNT -ne 0 ]; do
-    test -S $SOCKET && check_pgs
+    test -S $SOCKET && check_pgs && continue 2
     sleep 1
     let COUNT=COUNT-1
   done