mgr/cephadm: don't attempt daemon actions for daemons on offline hosts

They'll just fail anyway, and it will waste time waiting
for the connection to timeout. We have other places in
the serve loop that will check if the host is back
online.

Signed-off-by: Adam King <adking@redhat.com>
(cherry picked from commit e1d3e38f87bce4e53eb7e4c29c812a7ed750fa80)

diff --git a/src/pybind/mgr/cephadm/serve.py b/src/pybind/mgr/cephadm/serve.py
index b0060c75c2a6182d598475ce7d5af2c445e950c8..b969eea4e3195818846e119d603b2e9d698d06da 100644 (file)
--- a/src/pybind/mgr/cephadm/serve.py
+++ b/src/pybind/mgr/cephadm/serve.py
@@ -866,6 +866,12 @@ class CephadmServe:
             assert dd.hostname is not None
             assert dd.daemon_type is not None
             assert dd.daemon_id is not None
+
+            # any action we can try will fail for a daemon on an offline host,
+            # including removing the daemon
+            if dd.hostname in self.mgr.offline_hosts:
+                continue
+
             if not spec and dd.daemon_type not in ['mon', 'mgr', 'osd']:
                 # (mon and mgr specs should always exist; osds aren't matched
                 # to a service spec)