doc/cephadm: Add admonition re restarting an OSD service

Signed-off-by: Anthony D'Atri <anthonyeleven@users.noreply.github.com>

diff --git a/doc/cephadm/operations.rst b/doc/cephadm/operations.rst
index 22d91c39b062229dcdcdd6de6b020b4e7994be09..c6aeedd9d01a2ea5b456bb438618e92992c59c94 100644 (file)
--- a/doc/cephadm/operations.rst
+++ b/doc/cephadm/operations.rst
@@ -51,17 +51,23 @@ You can stop, start, or restart a daemon with:
 
 .. prompt:: bash #
 
-   ceph orch daemon stop <name>
-   ceph orch daemon start <name>
-   ceph orch daemon restart <name>
+   ceph orch daemon stop <daemonid>
+   ceph orch daemon start <daemonid>
+   ceph orch daemon restart <daemonid>
 
 You can also do the same for all daemons for a service with:   
 
 .. prompt:: bash #
 
-   ceph orch stop <name>
-   ceph orch start <name>
-   ceph orch restart <name>
+   ceph orch stop <serviceid>
+   ceph orch start <serviceid>
+   ceph orch restart <serviceid>
+
+.. note::
+    It is usually not safe to run ``ceph orch restart osd.myosdservice`` on a
+    running cluster, as attention is not paid to CRUSH failure domains, and
+    parallel OSD restarts may lead to temporary data unavailability or in rare
+    cases even data loss.
 
 
 Redeploying or reconfiguring a daemon

diff --git a/doc/cephadm/services/osd.rst b/doc/cephadm/services/osd.rst
index ed9b846052a1ebc4542dccd74130ae8864f2c95d..2dce4a99889a2ffeef1e85c41ade41a1ee3322e2 100644 (file)
--- a/doc/cephadm/services/osd.rst
+++ b/doc/cephadm/services/osd.rst
@@ -1215,6 +1215,13 @@ This example applies to two hosts: ``ceph01`` and ``ceph04``.
 post pertinent to its development can be seen here:*
 `Eugen Block's "Cephadm: Activate existing OSDs" blog post <https://heiterbiswolkig.blogs.nde.ag/2025/02/06/cephadm-activate-existing-osds/>`_.
 
+.. note::
+    It is usually not safe to run ``ceph orch restart osd.myosdservice`` on a
+    running cluster, as attention is not paid to CRUSH failure domains, and
+    parallel OSD restarts may lead to temporary data unavailability or in rare
+    cases even data loss.
+
+
 Further Reading
 ===============