From f72175fe29b5947b3276b042833da2cf8a5abe3b Mon Sep 17 00:00:00 2001
From: Thiago Miranda <ThiagoOliveira.Miranda@windriver.com>
Date: Mon, 27 Dec 2021 07:51:01 -0500
Subject: [PATCH] Fix cephfs access after upgrading

After the system upgrade, the ceph mds data was deleted, and we could
no longer find the files stored with CephFS.

This procedure resets and recreates the mds so that the files appear
again after the volume is mounted.

Test Plan:
 - Tested Upgrading procedure from 21.05 on AIO-SX
 - Tested B&R procedure on AIO-SX

Closes-bug: 1955823

Signed-off-by: Thiago Miranda <ThiagoOliveira.Miranda@windriver.com>
Change-Id: Ic146dda35c3c71cf11c73a4f972b274ec3404d07
---
 .../recover-ceph-data/files/recover_cephfs.sh     | 15 ++++++---------
 .../roles/recover-ceph-data/tasks/main.yml        |  3 +++
 2 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/playbookconfig/src/playbooks/roles/recover-ceph-data/files/recover_cephfs.sh b/playbookconfig/src/playbooks/roles/recover-ceph-data/files/recover_cephfs.sh
index df3f4e6a9..12aea97d3 100755
--- a/playbookconfig/src/playbooks/roles/recover-ceph-data/files/recover_cephfs.sh
+++ b/playbookconfig/src/playbooks/roles/recover-ceph-data/files/recover_cephfs.sh
@@ -31,15 +31,12 @@ set -x
 # Check if the filesystem for the system RWX provisioner is present
 ceph fs ls | grep ${FS_NAME}
 if [ $? -ne 0 ]; then
-    # Attempt to create the pool if not present, this should be present
-    ceph fs new ${FS_NAME} ${METADATA_POOL_NAME} ${DATA_POOL_NAME}
-    if [ $? -eq 22 ]; then
-        # We need to rebuild the fs since we have hit:
-        #   Error EINVAL: pool 'kube-cephfs-metadata' already contains some
-        #   objects. Use an empty pool instead.
-        ceph fs new ${FS_NAME} ${METADATA_POOL_NAME} ${DATA_POOL_NAME} --force
-        ceph fs reset ${FS_NAME} --yes-i-really-mean-it
-    fi
+    # If we have existing metadata/data pools, use them
+    ceph fs new ${FS_NAME} ${METADATA_POOL_NAME} ${DATA_POOL_NAME} --force
+    # Reset the filesystem and journal
+    ceph fs reset ${FS_NAME} --yes-i-really-mean-it
+    cephfs-journal-tool --rank=${FS_NAME}:0 event recover_dentries summary
+    cephfs-journal-tool --rank=${FS_NAME}:0 journal reset
 fi
 
 # Start the Ceph MDS
diff --git a/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml b/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml
index e8e6ca65c..6f1ef7824 100644
--- a/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml
+++ b/playbookconfig/src/playbooks/roles/recover-ceph-data/tasks/main.yml
@@ -281,6 +281,9 @@
       script: recover_cephfs.sh
       register: cephfs_recovery_out
 
+    - name: Create ceph.client.guest.keyring to allow ceph mount again
+      command: touch /etc/ceph/ceph.client.guest.keyring
+
     - debug: var=cephfs_recovery_out.stdout_lines
 
     - name: Restart ceph one more time to pick latest changes