cephadm: reset-failed on reconfig

Sometimes we had a bad config that reconfig will fix, so we need to
reset-failed before restart.  If there was not failure this is harmless.

Signed-off-by: Sage Weil <sage@redhat.com>

diff --git a/src/cephadm/cephadm b/src/cephadm/cephadm
index 47628f2b83eda8926eaa15a05fb85c644591ca57..59c7df807ffbb5ec05229fbf593e8613c9174015 100755 (executable)
--- a/src/cephadm/cephadm
+++ b/src/cephadm/cephadm
@@ -1259,6 +1259,8 @@ def deploy_daemon(fsid, daemon_type, daemon_id, c, uid, gid,
     if reconfig and daemon_type not in Ceph.daemons:
         # ceph daemons do not need a restart; others (presumably) do to pick
         # up the new config
+        call_throws(['systemctl', 'reset-failed',
+                     get_unit_name(fsid, daemon_type, daemon_id)])
         call_throws(['systemctl', 'restart',
                      get_unit_name(fsid, daemon_type, daemon_id)])