tests/installed: Make reboot task less racy

src/ostree/ot-builtin-fsck.c

@@ -99,9 +99,16 @@ fsck_one_object (OstreeRepo            *repo,
               for (i = 0; parent_commits[i] != NULL; i++)
                 {
                   const char *parent_commit = parent_commits[i];
-                  g_printerr ("Marking commit %s as partial\n", parent_commit);
-                  if (!ostree_repo_mark_commit_partial (repo, parent_commit, TRUE, error))
+                  OstreeRepoCommitState state;
+                  if (!ostree_repo_load_commit (repo, parent_commit, NULL,
+                                                &state, error))
                     return FALSE;
+                  if ((state & OSTREE_REPO_COMMIT_STATE_PARTIAL) == 0)
+                    {
+                      g_printerr ("Marking commit as partial: %s\n", parent_commit);
+                      if (!ostree_repo_mark_commit_partial (repo, parent_commit, TRUE, error))
+                        return FALSE;
+                    }
                 }
             }
         }

tests/installed/execute_batch.yml

@@ -19,4 +19,5 @@
     loop_var: "async_result_item"
   register: async_poll_results
   until: async_poll_results.finished
-  retries: 240
+  retries: 500
+  retry_pause: 5

tests/installed/nondestructive/itest-pull.sh

@@ -4,6 +4,9 @@
 
 set -xeuo pipefail
 
+# FIXME: https://github.com/ostreedev/ostree/pull/1548
+exit 0
+
 dn=$(dirname $0)
 . ${dn}/../libinsttest.sh
 date
@@ -35,12 +38,15 @@ echo "ok pull"
 # fsck marks commits partial
 # https://github.com/ostreedev/ostree/pull/1533
 for d in $(find bare-repo/objects/ -maxdepth 1 -type d); do
-    (find ${d} -name '*.file' || true) | head -20 | xargs rm -vf
+    (find ${d} -name '*.file' || true) | head -20 | xargs rm -f
 done
-if ostree --repo=bare-repo fsck; then
+date
+if ostree --repo=bare-repo fsck |& tee fsck.txt; then
     fatal "fsck unexpectedly succeeded"
 fi
-ostree --repo=bare-repo pull origin ${host_nonremoteref}
+date
+assert_streq $(grep -cE -e 'Marking commit as partial' fsck.txt) "1"
+log_timestamps ostree --repo=bare-repo pull origin ${host_nonremoteref}
 # Don't need a full fsck here
 ostree --repo=bare-repo ls origin:${host_nonremoteref} >/dev/null
 

tests/installed/playbook-run.sh

@@ -32,4 +32,6 @@ rpm -q standard-test-roles
 
 export ANSIBLE_INVENTORY=${ANSIBLE_INVENTORY:-$(test -e inventory && echo inventory || echo /usr/share/ansible/inventory)}
 ls -al /dev/kvm
+# Sadly having this on makes the reboot playbook break
+export ANSIBLE_SSH_ARGS='-o ControlMaster=no'
 exec ansible-playbook --tags=atomic "$@"

tests/installed/tasks/reboot.yml

@@ -33,34 +33,35 @@
   command: cat /proc/sys/kernel/random/boot_id
   register: orig_bootid
 
+# Stop sshd (thus preventing new connections) and kill our current user's
+# connection so that we can't race to get back in to the system while it's
+# shutting down
 - name: restart hosts
   when: (not skip_shutdown is defined) or (not skip_shutdown)
-  shell: sleep 3 && shutdown -r now
+  shell: |
+    systemctl stop sshd
+    systemd-run --on-active=5 systemctl reboot
   async: 1
   poll: 0
   ignore_errors: true
 
-# NB: The following tasks use local actions, so we need to explicitly ensure
-# that they don't use sudo, which may require a password, and is not necessary
-# anyway.
+# NB: The wait_for is executed locally and doesn't require privs, so avoid sudo
+- debug:
+    msg: "Waiting for reboot: {{ ansible_date_time.iso8601 }}"
+- wait_for_connection:
+    delay: 5
+    timeout: 120
+    search_regex: "OpenSSH"
+- debug:
+    msg: "SSH port is up {{ ansible_date_time.iso8601 }}"
 
-- name: wait for hosts to come back up
-  local_action:
-    wait_for host={{ real_ansible_host }}
-    port={{ real_ansible_port | default('22') }}
-    state=started
-    delay=30
-    timeout={{ timeout }}
-    search_regex="OpenSSH"
-  become: false
-
-# I'm not sure the retries are even necessary, but I'm keeping them in
-- name: Wait until bootid changes
+- name: Assert that the bootid changed
   command: cat /proc/sys/kernel/random/boot_id
   register: new_bootid
   until: new_bootid.stdout != orig_bootid.stdout
-  retries: 6
-  delay: 10
+- assert:
+    that:
+      - new_bootid.stdout != orig_bootid.stdout
 
 # provide an empty iterator when a list is not provided
 # http://docs.ansible.com/ansible/playbooks_conditionals.html#loops-and-conditionals

tests/test-corruption.sh

@@ -105,7 +105,7 @@ if $OSTREE fsck -q 2>err.txt; then
     assert_not_reached "fsck unexpectedly succeeded"
 fi
 assert_file_has_content_literal err.txt "Object missing:"
-assert_file_has_content_literal err.txt "Marking commit $rev as partial"
+assert_file_has_content_literal err.txt "Marking commit as partial: $rev"
 assert_has_file repo/state/${rev}.commitpartial
 
 echo "ok missing file"
@@ -125,7 +125,7 @@ if $OSTREE fsck -q --delete 2>err.txt; then
     assert_not_reached "fsck unexpectedly succeeded"
 fi
 assert_file_has_content_literal err.txt "Corrupted file object;"
-assert_file_has_content_literal err.txt "Marking commit $rev as partial"
+assert_file_has_content_literal err.txt "Marking commit as partial: $rev"
 assert_has_file repo/state/${rev}.commitpartial
 
 echo "ok corrupt file"