use Data::Dumper;
use Cwd qw(abs_path);
use Fcntl ':mode';
+use File::Basename;
use File::stat;
use JSON;
my $SGDISK = "/sbin/sgdisk";
my $PVS = "/sbin/pvs";
my $LVS = "/sbin/lvs";
-my $UDEVADM = "/bin/udevadm";
my $LSBLK = "/bin/lsblk";
+sub check_bin {
+ my ($path) = @_;
+ return -x $path;
+}
+
sub verify_blockdev_path {
my ($rel_path) = @_;
} elsif (defined($type) && $type eq 'text') {
$smartdata->{text} = '' if !defined $smartdata->{text};
$smartdata->{text} .= "$line\n";
- # extract wearout from nvme text, allow for decimal values
- if ($line =~ m/Percentage Used:\s*(\d+(?:\.\d+)?)\%/i) {
+ # extract wearout from nvme/sas text, allow for decimal values
+ if ($line =~ m/Percentage Used(?: endurance indicator)?:\s*(\d+(?:\.\d+)?)\%/i) {
$smartdata->{wearout} = 100 - $1;
}
} elsif ($line =~ m/SMART Disabled/) {
return $smartdata;
}
-sub get_parttype_info() {
- my $cmd = [$LSBLK, '--json', '-o', 'path,parttype'];
+sub get_lsblk_info() {
+ my $cmd = [$LSBLK, '--json', '-o', 'path,parttype,fstype'];
my $output = "";
my $res = {};
eval {
warn "$@\n" if $@;
my $list = $parsed->{blockdevices} // [];
- foreach my $dev (@$list) {
- next if !($dev->{parttype});
- my $type = $dev->{parttype};
- $res->{$type} = [] if !defined($res->{$type});
- push @{$res->{$type}}, $dev->{path};
- }
+ $res = { map {
+ $_->{path} => {
+ parttype => $_->{parttype},
+ fstype => $_->{fstype}
+ }
+ } @{$list} };
return $res;
}
my $get_devices_by_partuuid = sub {
- my ($parttype_map, $uuids, $res) = @_;
+ my ($lsblk_info, $uuids, $res) = @_;
$res = {} if !defined($res);
- foreach my $uuid (sort keys %$uuids) {
- map { $res->{$_} = $uuids->{$uuid} } @{$parttype_map->{$uuid}};
+ foreach my $dev (sort keys %{$lsblk_info}) {
+ my $uuid = $lsblk_info->{$dev}->{parttype};
+ next if !defined($uuid) || !defined($uuids->{$uuid});
+ $res->{$dev} = $uuids->{$uuid};
}
return $res;
};
sub get_zfs_devices {
- my ($parttype_map) = @_;
+ my ($lsblk_info) = @_;
my $res = {};
- return {} if ! -x $ZPOOL;
+ return {} if !check_bin($ZPOOL);
# use zpool and parttype uuid,
# because log and cache do not have
};
- $res = $get_devices_by_partuuid->($parttype_map, $uuids, $res);
+ $res = $get_devices_by_partuuid->($lsblk_info, $uuids, $res);
return $res;
}
sub get_lvm_devices {
- my ($parttype_map) = @_;
+ my ($lsblk_info) = @_;
my $res = {};
eval {
run_command([$PVS, '--noheadings', '--readonly', '-o', 'pv_name'], outfunc => sub{
};
# if something goes wrong, we do not want
- # to give up, but indicate an error has occured
+ # to give up, but indicate an error has occurred
warn "$@\n" if $@;
my $uuids = {
"e6d6d379-f507-44c2-a23c-238f2a3df928" => 1,
};
- $res = $get_devices_by_partuuid->($parttype_map, $uuids, $res);
+ $res = $get_devices_by_partuuid->($lsblk_info, $uuids, $res);
return $res;
}
sub get_ceph_journals {
- my ($parttype_map) = @_;
+ my ($lsblk_info) = @_;
my $res = {};
my $uuids = {
'cafecafe-9b03-4f30-b4c6-b4b80ceff106' => 4, # block
};
- $res = $get_devices_by_partuuid->($parttype_map, $uuids, $res);
+ $res = $get_devices_by_partuuid->($lsblk_info, $uuids, $res);
return $res;
}
my $fields = [ split(';', $line) ];
# lvs syntax is /dev/sdX(Y) where Y is the start (which we do not need)
- my ($dev) = $fields->[0] =~ m|^(/dev/[a-z]+)|;
+ my ($dev) = $fields->[0] =~ m|^(/dev/[a-z]+[^(]*)|;
if ($fields->[1] =~ m|^osd-([^-]+)-|) {
my $type = $1;
# $result autovivification is wanted, to not creating empty hashes
my $info = "";
my $data = {};
eval {
- run_command([$UDEVADM, 'info', '-p', $dev, '--query', 'all'], outfunc => sub {
+ run_command(['udevadm', 'info', '-p', $dev, '--query', 'all'], outfunc => sub {
my ($line) = @_;
$info .= "$line\n";
});
return $data;
}
+sub get_sysdir_size {
+ my ($sysdir) = @_;
+
+ my $size = file_read_firstline("$sysdir/size");
+ return if !$size;
+
+ # linux always considers sectors to be 512 bytes,
+ # independently of real block size
+ return $size * 512;
+}
+
sub get_sysdir_info {
my ($sysdir) = @_;
my $data = {};
- my $size = file_read_firstline("$sysdir/size");
- return undef if !$size;
-
- # linux always considers sectors to be 512 bytes,
- # independently of real block size
- $data->{size} = $size * 512;
+ $data->{size} = get_sysdir_size($sysdir) or return;
# dir/queue/rotational should be 1 for hdd, 0 for ssd
$data->{rotational} = file_read_firstline("$sysdir/queue/rotational") // -1;
return $type eq 'ssd' || $type eq 'nvme';
}
-sub get_disks {
- my ($disks, $nosmart) = @_;
- my $disklist = {};
-
+sub mounted_blockdevs {
my $mounted = {};
my $mounts = PVE::ProcFSTools::parse_proc_mounts();
$mounted->{abs_path($mount->[0])} = $mount->[1];
};
- my $dev_is_mounted = sub {
- my ($dev) = @_;
- return $mounted->{$dev};
- };
+ return $mounted;
+}
+
+sub get_disks {
+ my ($disks, $nosmart, $include_partitions) = @_;
+ my $disklist = {};
- my $parttype_map = get_parttype_info();
+ my $mounted = mounted_blockdevs();
- my $journalhash = get_ceph_journals($parttype_map);
+ my $lsblk_info = get_lsblk_info();
+
+ my $journalhash = get_ceph_journals($lsblk_info);
my $ceph_volume_infos = get_ceph_volume_infos();
- my $zfshash = get_zfs_devices($parttype_map);
+ my $zfshash = get_zfs_devices($lsblk_info);
- my $lvmhash = get_lvm_devices($parttype_map);
+ my $lvmhash = get_lvm_devices($lsblk_info);
my $disk_regex = ".*";
if (defined($disks)) {
die "disks is not a string or array reference\n";
}
# we get cciss/c0d0 but need cciss!c0d0
- map { s|cciss/|cciss!| } @$disks;
+ $_ =~ s|cciss/|cciss!| for @$disks;
$disk_regex = "(?:" . join('|', @$disks) . ")";
}
};
}
- my $used;
-
- $used = 'LVM' if $lvmhash->{$devpath};
-
- $used = 'mounted' if &$dev_is_mounted($devpath);
-
- $used = 'ZFS' if $zfshash->{$devpath};
-
# we replaced cciss/ with cciss! above
# but in the result we need cciss/ again
# because the caller might want to check the
my $db_count = 0;
my $wal_count = 0;
- my $found_partitions;
- my $found_lvm;
- my $found_mountpoints;
- my $found_zfs;
- my $found_dm;
my $partpath = $devpath;
# remove part after last / to
# e.g. from /dev/cciss/c0d0 get /dev/cciss
$partpath =~ s/\/[^\/]+$//;
+ my $determine_usage = sub {
+ my ($devpath, $sysdir, $is_partition) = @_;
+
+ return 'LVM' if $lvmhash->{$devpath};
+ return 'ZFS' if $zfshash->{$devpath};
+
+ my $info = $lsblk_info->{$devpath} // {};
+
+ my $parttype = $info->{parttype};
+ if (defined($parttype)) {
+ return 'BIOS boot'
+ if $parttype eq '21686148-6449-6e6f-744e-656564454649';
+ return 'EFI'
+ if $parttype eq 'c12a7328-f81f-11d2-ba4b-00a0c93ec93b';
+ return 'ZFS reserved'
+ if $parttype eq '6a945a3b-1dd2-11b2-99a6-080020736631';
+ }
+
+ my $fstype = $info->{fstype};
+ if (defined($fstype)) {
+ return "${fstype} (mounted)" if $mounted->{$devpath};
+ return "${fstype}";
+ }
+ return 'mounted' if $mounted->{$devpath};
+
+ return if !$is_partition;
+
+ # for devices, this check is done explicitly later
+ return 'Device Mapper' if !dir_is_empty("$sysdir/holders");
+
+ return 'partition';
+ };
+
+ my $collect_ceph_info = sub {
+ my ($devpath) = @_;
+
+ my $ceph_volume = $ceph_volume_infos->{$devpath} or return;
+ $journal_count += $ceph_volume->{journal} // 0;
+ $db_count += $ceph_volume->{db} // 0;
+ $wal_count += $ceph_volume->{wal} // 0;
+ if (defined($ceph_volume->{osdid})) {
+ $osdid = $ceph_volume->{osdid};
+ $bluestore = 1 if $ceph_volume->{bluestore};
+ $osdencrypted = 1 if $ceph_volume->{encrypted};
+ }
+
+ my $result = { %{$ceph_volume} };
+ $result->{journals} = delete $result->{journal}
+ if $result->{journal};
+ return $result;
+ };
+
+ my $partitions = {};
+
dir_glob_foreach("$sysdir", "$dev.+", sub {
my ($part) = @_;
- $found_partitions = 1;
+ $partitions->{$part} = $collect_ceph_info->("$partpath/$part");
+ my $lvm_based_osd = defined($partitions->{$part});
+
+ $partitions->{$part}->{devpath} = "$partpath/$part";
+ $partitions->{$part}->{parent} = "$devpath";
+ $partitions->{$part}->{gpt} = $data->{gpt};
+ $partitions->{$part}->{type} = 'partition';
+ $partitions->{$part}->{size} =
+ get_sysdir_size("$sysdir/$part") // 0;
+ $partitions->{$part}->{used} =
+ $determine_usage->("$partpath/$part", "$sysdir/$part", 1);
+ $partitions->{$part}->{osdid} //= -1;
+
+ # Avoid counting twice (e.g. partition on which the LVM for the
+ # DB OSD resides is present in the $journalhash)
+ return if $lvm_based_osd;
- if (my $mp = &$dev_is_mounted("$partpath/$part")) {
- $found_mountpoints = 1;
+ # Legacy handling for non-LVM based OSDs
+
+ if (my $mp = $mounted->{"$partpath/$part"}) {
if ($mp =~ m|^/var/lib/ceph/osd/ceph-(\d+)$|) {
$osdid = $1;
+ $partitions->{$part}->{osdid} = $osdid;
}
}
- if ($lvmhash->{"$partpath/$part"}) {
- $found_lvm = 1;
- }
-
- if ($zfshash->{"$partpath/$part"}) {
- $found_zfs = 1;
- }
-
if (my $journal_part = $journalhash->{"$partpath/$part"}) {
$journal_count++ if $journal_part == 1;
$db_count++ if $journal_part == 2;
$wal_count++ if $journal_part == 3;
$bluestore = 1 if $journal_part == 4;
- }
- if (!dir_is_empty("$sysdir/$part/holders") && !$found_lvm) {
- $found_dm = 1;
+ $partitions->{$part}->{journals} = 1 if $journal_part == 1;
+ $partitions->{$part}->{db} = 1 if $journal_part == 2;
+ $partitions->{$part}->{wal} = 1 if $journal_part == 3;
+ $partitions->{$part}->{bluestore} = 1 if $journal_part == 4;
}
});
- if (my $ceph_volume = $ceph_volume_infos->{$devpath}) {
- $journal_count += $ceph_volume->{journal} // 0;
- $db_count += $ceph_volume->{db} // 0;
- $wal_count += $ceph_volume->{wal} // 0;
- if (defined($ceph_volume->{osdid})) {
- $osdid = $ceph_volume->{osdid};
- $bluestore = 1 if $ceph_volume->{bluestore};
- $osdencrypted = 1 if $ceph_volume->{encrypted};
+ my $used = $determine_usage->($devpath, $sysdir, 0);
+ if (!$include_partitions) {
+ foreach my $part (sort keys %{$partitions}) {
+ next if $partitions->{$part}->{used} eq 'partition';
+ $used //= $partitions->{$part}->{used};
}
+ } else {
+ # fstype might be set even if there are partitions, but showing that is confusing
+ $used = 'partitions' if scalar(keys %{$partitions});
}
-
- $used = 'mounted' if $found_mountpoints && !$used;
- $used = 'LVM' if $found_lvm && !$used;
- $used = 'ZFS' if $found_zfs && !$used;
- $used = 'Device Mapper' if $found_dm && !$used;
- $used = 'partitions' if $found_partitions && !$used;
-
+ $used //= 'partitions' if scalar(keys %{$partitions});
# multipath, software raid, etc.
# this check comes in last, to show more specific info
# if we have it
- $used = 'Device Mapper' if !$used && !dir_is_empty("$sysdir/holders");
+ $used //= 'Device Mapper' if !dir_is_empty("$sysdir/holders");
$disklist->{$dev}->{used} = $used if $used;
+
+ $collect_ceph_info->($devpath);
+
$disklist->{$dev}->{osdid} = $osdid;
$disklist->{$dev}->{journals} = $journal_count if $journal_count;
$disklist->{$dev}->{bluestore} = $bluestore if $osdid != -1;
$disklist->{$dev}->{osdencrypted} = $osdencrypted if $osdid != -1;
$disklist->{$dev}->{db} = $db_count if $db_count;
$disklist->{$dev}->{wal} = $wal_count if $wal_count;
+
+ if ($include_partitions) {
+ foreach my $part (keys %{$partitions}) {
+ $disklist->{$part} = $partitions->{$part};
+ }
+ }
});
return $disklist;
my $st = stat($part_path);
- next if !$st->mode || !S_ISBLK($st->mode) || !$st->rdev;
+ die "error detecting block device '$part_path'\n"
+ if !$st || !$st->mode || !S_ISBLK($st->mode) || !$st->rdev;
+
my $major = PVE::Tools::dev_t_major($st->rdev);
my $minor = PVE::Tools::dev_t_minor($st->rdev);
my $partnum_path = "/sys/dev/block/$major:$minor/";
die "No valid block device\n" if index($dev, $block_dev) == -1;
$block_dev = "/dev/$block_dev";
- die "Block device does not exsists\n" if !(-b $block_dev);
+ die "Block device does not exists\n" if !(-b $block_dev);
return $block_dev;
}
my $partition;
- # loop again to detect the real partiton device which does not always follow
+ # loop again to detect the real partition device which does not always follow
# a strict $devname$partition scheme like /dev/nvme0n1 -> /dev/nvme0n1p1
dir_glob_foreach("/sys/block/$devname", qr/\Q$devname\E.*$newpartid/, sub {
my ($part) = @_;
return $partition;
}
+my sub strip_dev :prototype($) {
+ my ($devpath) = @_;
+ $devpath =~ s|^/dev/||;
+ return $devpath;
+}
+
+# Check if a disk or any of its partitions has a holder.
+# Can also be called with a partition.
+# Expected to be called with a result of verify_blockdev_path().
+sub has_holder {
+ my ($devpath) = @_;
+
+ my $dev = strip_dev($devpath);
+
+ return $devpath if !dir_is_empty("/sys/class/block/${dev}/holders");
+
+ my $found;
+ dir_glob_foreach("/sys/block/${dev}", "${dev}.+", sub {
+ my ($part) = @_;
+ $found = "/dev/${part}" if !dir_is_empty("/sys/class/block/${part}/holders");
+ });
+
+ return $found;
+}
+
+# Basic check if a disk or any of its partitions is mounted.
+# Can also be called with a partition.
+# Expected to be called with a result of verify_blockdev_path().
+sub is_mounted {
+ my ($devpath) = @_;
+
+ my $mounted = mounted_blockdevs();
+
+ return $devpath if $mounted->{$devpath};
+
+ my $dev = strip_dev($devpath);
+
+ my $found;
+ dir_glob_foreach("/sys/block/${dev}", "${dev}.+", sub {
+ my ($part) = @_;
+ my $partpath = "/dev/${part}";
+
+ $found = $partpath if $mounted->{$partpath};
+ });
+
+ return $found;
+}
+
+# Wipes all labels and the first 200 MiB of a disk/partition (or the whole if it is smaller).
+# Expected to be called with a result of verify_blockdev_path().
+sub wipe_blockdev {
+ my ($devpath) = @_;
+
+ 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 "Couldn'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;
+
+ my $to_wipe = [];
+ dir_glob_foreach("/sys/class/block/${devname}", "${devname}.+", sub {
+ my ($part) = @_;
+ push $to_wipe->@*, "/dev/${part}" if -b "/dev/${part}";
+ });
+
+ if (scalar($to_wipe->@*) > 0) {
+ print "found child partitions to wipe: ". join(', ', $to_wipe->@*) ."\n";
+ }
+ push $to_wipe->@*, $devpath; # put actual device last
+
+ print "wiping block device ${devpath}\n";
+
+ run_command(['wipefs', '--all', $to_wipe->@*], errmsg => "error wiping '${devpath}'");
+
+ run_command(
+ ['dd', 'if=/dev/zero', "of=${devpath}", 'bs=1M', 'conv=fdatasync', "count=${count}"],
+ errmsg => "error wiping '${devpath}'",
+ );
+}
+
1;