From c2a93974742c24ab0fcdcd1651a3b2dde63ddf1c Mon Sep 17 00:00:00 2001
From: Mike Christie <mchristi@redhat.com>
Date: Tue, 28 Jan 2020 16:31:55 -0600
Subject: [PATCH] iscsi: Fix crashes during rolling update

During a rolling update we will run the ceph iscsigw tasks that start
the daemons then run the configure_iscsi.yml tasks which can create
iscsi objects like targets, disks, clients, etc. The problem is that
once the daemons are started they will accept confifguration requests,
or may want to update the system themself. Those operations can then
conflict with the configure_iscsi.yml tasks that setup objects and we
can end up in crashes due to the kernel being in a unsupported state.

This could also happen during creation, but is less likely due to no
objects being setup yet, so there are no watchers or users accessing the
gws yet. The fix in this patch works for both update and initial setup.

Resolves: https://bugzilla.redhat.com/show_bug.cgi?id=1795806

Signed-off-by: Mike Christie <mchristi@redhat.com>
(cherry picked from commit 77f3b5d51b84a6338847c5f6a93f22a3a6a683d2)
---
 roles/ceph-iscsi-gw/tasks/main.yml                |  4 ++++
 .../tasks/non-container/postrequisites.yml        |  9 +++++++++
 .../tasks/non-container/prerequisites.yml         | 15 +++++++--------
 3 files changed, 20 insertions(+), 8 deletions(-)
 create mode 100644 roles/ceph-iscsi-gw/tasks/non-container/postrequisites.yml

diff --git a/roles/ceph-iscsi-gw/tasks/main.yml b/roles/ceph-iscsi-gw/tasks/main.yml
index 77cfe618e..297eb4d88 100644
--- a/roles/ceph-iscsi-gw/tasks/main.yml
+++ b/roles/ceph-iscsi-gw/tasks/main.yml
@@ -20,6 +20,10 @@
   when:
     - not containerized_deployment
 
+- name: include non-container/postrequisites.yml
+  include_tasks: non-container/postrequisites.yml
+  when: not containerized_deployment | bool
+
 - name: include containerized.yml
   include_tasks: container/containerized.yml
   when:
diff --git a/roles/ceph-iscsi-gw/tasks/non-container/postrequisites.yml b/roles/ceph-iscsi-gw/tasks/non-container/postrequisites.yml
new file mode 100644
index 000000000..5242525ef
--- /dev/null
+++ b/roles/ceph-iscsi-gw/tasks/non-container/postrequisites.yml
@@ -0,0 +1,9 @@
+- name: start rbd-target-api and rbd-target-gw
+  service:
+    name: "{{ item }}"
+    state: started
+    enabled: yes
+    masked: no
+  with_items:
+    - rbd-target-gw
+    - rbd-target-api
diff --git a/roles/ceph-iscsi-gw/tasks/non-container/prerequisites.yml b/roles/ceph-iscsi-gw/tasks/non-container/prerequisites.yml
index f3ed7e738..56e69be0f 100644
--- a/roles/ceph-iscsi-gw/tasks/non-container/prerequisites.yml
+++ b/roles/ceph-iscsi-gw/tasks/non-container/prerequisites.yml
@@ -89,14 +89,13 @@
     - target.stat.exists
     - target.stat.islnk == False
 
-- name: enable the rbd-target-gw service and make sure it is running
+# Only start tcmu-runner, so configure_iscsi.yml can create disks.
+# We must start rbd-target-gw/api after configure_iscsi.yml to avoid
+# races where they are both trying to setup the same object during
+# a rolling update.
+- name: start tcmu-runner
   service:
-    name: rbd-target-gw
-    enabled: yes
+    name: tcmu-runner
     state: started
-
-- name: enable the rbd-target-api service and make sure it is running
-  service:
-    name: rbd-target-api
     enabled: yes
-    state: started
+    masked: no