FIXUP 94834d6

Signed-off-by: Wesley Hershberger <[email protected]>

doc/howto/cluster_recover.md

@@ -9,9 +9,14 @@ If you can bring the offline cluster members back or delete them from the cluste
 If the cluster members are lost permanently (e.g. disk failure), it is possible to recover them if at least one member survives.
 
 ```{note}
-When your cluster is in a state that needs recovery, most `lxc` commands do not work, because the LXD daemon cannot connect to the LXD database.
+When your cluster is in a state that needs recovery, most `lxc` commands do not
+work because the LXD database does not respond when a majority of database
+voters are inaccessible.
+
+The commands to recover a cluster are provided directly by the LXD daemon (`lxd`)
+because they modify database files directly instead of making requests to the
+LXD daemon.
 
-The commands to recover a cluster are provided directly by the LXD daemon (`lxd`).
 Run `lxd cluster --help` for an overview of all available commands.
 ```
 
@@ -26,7 +31,9 @@ To determine which members have (or had) database roles, log on to any surviving
 
 ## Recover from quorum loss
 
-If only one cluster member with the database role survives, complete the following steps:
+If only one cluster member with the database role survives, complete the following
+steps. See [Reconfigure the cluster](#reconfigure-the-cluster) below for recovering
+more than one member.
 
 1. Make sure that the LXD daemon is not running on the machine.
    For example, if you're using the snap:
@@ -99,9 +106,9 @@ Complete the following steps on one database member:
    You can edit the addresses and the roles.
 
 1. When the cluster configuration has been changed on one member, LXD will create
-   a tarball of the global database at `/var/snap/lxd/common/lxd/database/recovery_db.tar.gz`.
-   Copy this recovery tarball to the database directory (`/var/snap/lxd/common/lxd/database`)
-  on all remaining cluster members.
+   a tarball of the global database (`/var/snap/lxd/common/lxd/database/recovery_db.tar.gz`
+   for snap installations or `/var/lib/lxd/database/recovery_db.tar.gz`).
+   Copy this recovery tarball to the same path on all remaining cluster members.
 
 1. Once the tarball has been copied to all remaining cluster members, start the
    LXD daemon on all members again. LXD will load the recovery tarball on startup.