Set ceph osd max object name and namespace len on upgrade when on ext4

As per [1] we need to lower osd max object name and namespace len when
upgrading from Hammer and the OSD is backed by ext4.

These could also be given via ExtraConfig but on upgrade we only run
puppet apply after this script is executed, so the values won't be
effective unless the daemon is restarted. Yet we do not want puppet
to restart the daemon because we can't bring all OSDs down
unconditionally or guests will die.

1. http://tracker.ceph.com/issues/16187

Co-Authored-By: Michele Baldessari <michele@acksyn.org>
Co-Authored-By: Dimitri Savineau <dsavinea@redhat.com>
Change-Id: I7fec4e2426bdacd5f364adbebd42ab23dcfa523a
Closes-Bug: 1628874

diff --git a/extraconfig/tasks/major_upgrade_ceph_storage.sh b/extraconfig/tasks/major_upgrade_ceph_storage.sh
index 705e84e..56b54e2 100644 (file)
--- a/extraconfig/tasks/major_upgrade_ceph_storage.sh
+++ b/extraconfig/tasks/major_upgrade_ceph_storage.sh
@@ -69,6 +69,16 @@ elif [[ "$UPDATED_VERSION" =~ ^10\.2.* ]]; then
     # Replay udev events with newer rules
     udevadm trigger && udevadm settle
 
+    # If on ext4, we need to enforce lower values for name and namespace len
+    # or ceph-osd will refuse to start, see: http://tracker.ceph.com/issues/16187
+    for OSD_ID in $OSD_IDS; do
+      OSD_FS=$(findmnt -n -o FSTYPE -T /var/lib/ceph/osd/ceph-${OSD_ID})
+      if [ ${OSD_FS} = ext4 ]; then
+        crudini --set /etc/ceph/ceph.conf global osd_max_object_name_len 256
+        crudini --set /etc/ceph/ceph.conf global osd_max_object_namespace_len 64
+      fi
+    done
+
     # Enable systemd unit
     systemctl enable ceph-osd.target
     for OSD_ID in $OSD_IDS; do