[pve-devel] [PATCH manager v4 1/2] Fix #2051: preserve DB/WAL disk on destroy

[Alwin Antreich]

'pveceph osd destroy <num> --cleanup'

When executing the command above, all disks associated with the OSD are
at the moment wiped with dd (incl. separate disks with DB/WAL).

The patch adds the ability to 'wipe_disks' to wipe the partition instead
of the whole disk.

Signed-off-by: Alwin Antreich <a.antreich at proxmox.com>
---
V3 -> V4:
    - incorporate suggestion by Thomas
    - and add second patch for cleanup
    https://pve.proxmox.com/pipermail/pve-devel/2019-February/035573.html

 PVE/API2/Ceph/OSD.pm | 15 ++++++++-------
 PVE/Ceph/Tools.pm    | 19 +++++++++++++++----
 2 files changed, 23 insertions(+), 11 deletions(-)

diff --git a/PVE/API2/Ceph/OSD.pm b/PVE/API2/Ceph/OSD.pm
index b4dc277e..c3858831 100644
--- a/PVE/API2/Ceph/OSD.pm
+++ b/PVE/API2/Ceph/OSD.pm
@@ -17,6 +17,7 @@ use PVE::RADOS;
 use PVE::RESTHandler;
 use PVE::RPCEnvironment;
 use PVE::Tools qw(run_command file_set_contents);
+use PVE::ProcFSTools;
 
 use base qw(PVE::RESTHandler);
 
@@ -394,7 +395,6 @@ __PACKAGE__->register_method ({
 	    # try to unmount from standard mount point
 	    my $mountpoint = "/var/lib/ceph/osd/ceph-$osdid";
 
-	    my $disks_to_wipe = {};
 	    my $remove_partition = sub {
 		my ($part) = @_;
 
@@ -402,11 +402,10 @@ __PACKAGE__->register_method ({
 		my $partnum = PVE::Diskmanage::get_partnum($part);
 		my $devpath = PVE::Diskmanage::get_blockdev($part);
 
+		PVE::Ceph::Tools::wipe_disks($part);
 		print "remove partition $part (disk '${devpath}', partnum $partnum)\n";
 		eval { run_command(['/sbin/sgdisk', '-d', $partnum, "${devpath}"]); };
 		warn $@ if $@;
-
-		$disks_to_wipe->{$devpath} = 1;
 	    };
 
 	    my $partitions_to_remove = [];
@@ -418,7 +417,9 @@ __PACKAGE__->register_method ({
 			next if !($dev && $path && $fstype);
 			next if $dev !~ m|^/dev/|;
 			if ($path eq $mountpoint) {
-			    my $data_part = abs_path($dev);
+			    my ($data_part) = abs_path($dev) =~ m|^(/.+)|
+				or die "invalid path: $path \n"; # untaint $part
+
 			    push @$partitions_to_remove, $data_part;
 			    last;
 			}
@@ -427,7 +428,9 @@ __PACKAGE__->register_method ({
 		}
 
 		foreach my $path (qw(journal block block.db block.wal)) {
-		    my $part = abs_path("$mountpoint/$path");
+		    my ($part) = abs_path("$mountpoint/$path") =~ m|^(/.+)|
+			or die "invalid path: $path \n"; # untaint $part
+
 		    if ($part) {
 			push @$partitions_to_remove, $part;
 		    }
@@ -443,8 +446,6 @@ __PACKAGE__->register_method ({
 		foreach my $part (@$partitions_to_remove) {
 		    $remove_partition->($part);
 		}
-
-		PVE::Ceph::Tools::wipe_disks(keys %$disks_to_wipe);
 	    }
 	};
 
diff --git a/PVE/Ceph/Tools.pm b/PVE/Ceph/Tools.pm
index 0ada98cf..9b02d7c8 100644
--- a/PVE/Ceph/Tools.pm
+++ b/PVE/Ceph/Tools.pm
@@ -4,6 +4,7 @@ use strict;
 use warnings;
 
 use File::Path;
+use File::Basename;
 use IO::File;
 
 use PVE::Tools qw(run_command dir_glob_foreach);
@@ -232,11 +233,21 @@ sub systemd_managed {
 sub wipe_disks {
     my (@devs) = @_;
 
-    my @wipe_cmd = qw(/bin/dd if=/dev/zero bs=1M count=200 conv=fdatasync);
+    my @wipe_cmd = qw(/bin/dd if=/dev/zero bs=1M conv=fdatasync);
+
     foreach my $devpath (@devs) {
-        print "wipe disk: $devpath\n";
-        eval { run_command([@wipe_cmd, "of=${devpath}"]) };
-        warn $@ if $@;
+	my $devname = basename($devpath);
+	my $dev_size = PVE::Tools::file_get_contents("/sys/class/block/$devname/size");
+
+	($dev_size) = $dev_size =~ m|(\d+)|; # untaint $dev_size
+	die "Coulnd't get the size of the device $devname\n" if (!defined($dev_size));
+
+	my $size = ($dev_size * 512 / 1024 / 1024);
+	my $count = ($size < 200) ? $size : 200;
+
+	print "wipe disk/partition: $devpath\n";
+	eval { run_command([@wipe_cmd, "count=$count", "of=${devpath}"]) };
+	warn $@ if $@;
     }
 };
 
-- 
2.11.0