[PATCH cluster v4 2/2] pmxcfs: don't abort join when backup fails and keep old config database

[Shannon Sterz <s.sterz@proxmox.com>]

by moving it out of the way of the new database instead of unlinking
it. this should improve situations where a failed backup would leave
the cluster that is being joined in an inconsistent state, while
making sure that the pre-existing database on the joining node stays
around.

Signed-off-by: Shannon Sterz <s.sterz@proxmox.com>
---
 src/PVE/Cluster.pm       | 5 +++++
 src/PVE/Cluster/Setup.pm | 5 +++--
 2 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/src/PVE/Cluster.pm b/src/PVE/Cluster.pm
index e96a7fe..e678479 100644
--- a/src/PVE/Cluster.pm
+++ b/src/PVE/Cluster.pm
@@ -917,4 +917,9 @@ sub cfs_backup_database {
     return $dbfile;
 }
 
+# NOTE: Should only be called on cluster join to clean up a database pre-join.
+sub cfs_move_existing_db {
+    rename $dbfile, "$dbfile.bak";
+}
+
 1;
diff --git a/src/PVE/Cluster/Setup.pm b/src/PVE/Cluster/Setup.pm
index 75d3507..9de6235 100644
--- a/src/PVE/Cluster/Setup.pm
+++ b/src/PVE/Cluster/Setup.pm
@@ -808,8 +808,9 @@ sub finish_join {
     my $cmd = ['systemctl', 'stop', 'pve-cluster'];
     PVE::Tools::run_command($cmd, errmsg => "can't stop pve-cluster service");
 
-    my $dbfile = PVE::Cluster::cfs_backup_database();
-    unlink $dbfile;
+    eval { PVE::Cluster::cfs_backup_database(); };
+    warn $@ if $@;
+    PVE::Cluster::cfs_move_existing_db();
 
     $cmd = ['systemctl', 'start', 'corosync', 'pve-cluster'];
     PVE::Tools::run_command($cmd, errmsg => "starting pve-cluster failed");
-- 
2.47.3