add untracked files

collectd/usr/local/bin/btrfs-data

@@ -0,0 +1,212 @@
+#!/usr/bin/python -u
+
+#
+# Imports
+#
+import sys
+import time
+import commands
+import argparse
+
+
+#
+# Misc
+#
+#sys.tracebacklimit = 0
+
+
+#
+# Global variables
+#
+size_data_total = 0
+size_data_exclusive = 0
+size_snapshot_total = 0
+size_snapshot_exclusive = 0
+
+
+#
+# Methods
+#
+def get_subvol_list(path):
+    command = 'sudo btrfs subvolume list -t %s' % (path)
+    status, output = commands.getstatusoutput(command)
+
+    if status is not 0:
+        raise Exception(command)
+
+    # Every line contains the following values: subvol_id, gen, toplevel, path
+    return output.splitlines()[2:]
+
+def get_filesystem_size(path):
+    command = 'sudo btrfs filesystem show --raw %s' % (path)
+    status, output = commands.getstatusoutput(command)
+
+    if status is not 0 or True:
+        # This command fails when running inside Docker container
+        # return maximum size of any filesystem instead
+        command = 'sudo btrfs filesystem show --raw'
+        status, output = commands.getstatusoutput(command)
+        lines = output.splitlines()
+        lines = filter(lambda x: 'devid' in x, lines)
+        sizes = [int(line.split()[3]) for line in lines]
+        return max(sizes)
+
+    # The sizes are on the third line
+    line = output.splitlines()[2]
+
+    # Element 3 and 5 respectively contain total and used sizes
+    return int(line.split()[3])
+
+def get_id_root(name, path):
+    lines = get_subvol_list(path)
+
+    # Filter lines where toplevel == 5
+    subvol_ids = filter(lambda x: int(x.split()[2]) == 5, lines)
+
+    # Try to retrieve the subvol_id for the root subvolume (if any)
+    if len(subvol_ids) == 1:
+        # The path contains a btrfs filesystem without subvolume for data
+        return int(subvol_ids[0].split()[0])
+    else:
+        # The path contains a btrfs filesystem with multiple subvolumes for data
+        try:
+            return int(filter(lambda x: x.split()[3] == name, subvol_ids)[0].split()[0])
+        except IndexError:
+            pass
+            
+
+    # Volume not found, root is probably the btrfs default (5)
+    return 5
+            
+def get_id_subvolumes(path, subvol_id):
+    lines = get_subvol_list(path)
+    lines = filter(lambda x: int(x.split()[2]) == subvol_id, lines)
+    return list(map(lambda x: int(x.split()[0]), lines))
+
+
+def get_disk_usage(name, path):
+    id_root = get_id_root(name, path)
+    id_subvolumes = get_id_subvolumes(path, id_root)
+    size_filesystem = get_filesystem_size(path)
+
+    # Get disk usage from quota
+    command = 'sudo btrfs qgroup show --raw %s' % (path)
+    status, output = commands.getstatusoutput(command)
+
+    if status is not 0:
+        raise Exception(command)
+
+    lines = output.splitlines()[2:]
+
+    # Global variables
+    global size_data_total
+    global size_data_exclusive
+    global size_snapshot_total
+    global size_snapshot_exclusive
+
+    # Total data volume in subvolume
+    size_data_total = 0
+
+    # Total data volume in snapshots
+    #   -> this variable is useless
+    size_snapshot_total = 0
+
+    # Data exclusively in subvolume
+    #   -> data that is not (yet) incorporated in a snapshot
+    size_data_exclusive = 0
+
+    # Data exclusively available in snapshots
+    #   -> data that was removed from volume
+    size_snapshot_exclusive = 0
+
+    for line in lines:
+        split = line.split()
+        subvol_id = 0
+        size_total = 0
+        size_exclusive = 0
+        try:
+            subvol_id = int(split[0].split('/')[1])
+            size_total = float(split[1])
+            size_exclusive = float(split[2])
+        except IndexError:
+            # ignore 'WARNING: Quota disabled'
+            pass
+
+        # size_exclusive is incorrect when snapshot is
+        # removed and qgroups are not updated yet,
+        # ignore the value when it seems unrealistic
+        if size_exclusive > size_filesystem:
+            size_exclusive = 0
+
+        if subvol_id == id_root:
+            size_data_total = size_total
+            size_data_exclusive = size_exclusive
+        elif subvol_id in id_subvolumes:
+            size_snapshot_total += size_total
+            size_snapshot_exclusive += size_exclusive
+
+def rescan_quota(path):
+    command = 'sudo btrfs quota rescan %s' % (path)
+    status, output = commands.getstatusoutput(command)
+    if status is not 0:
+        Exception(command)
+
+def print_human_readable(name):
+    global size_data_total
+    global size_data_exclusive
+    global size_snapshot_exclusive
+    size_data_total = size_data_total / (1024*1e6)
+    size_data_exclusive = size_data_exclusive / (1024*1e6)
+    size_snapshot_exclusive = size_snapshot_exclusive / (1024*1e6)
+    print '%10s: %6.1f Gb, %6.1f Gb, %6.1f Gb' % (name, size_data_total, size_data_exclusive, size_snapshot_exclusive)
+
+def print_rrd(name):
+    timestamp = int(time.time())
+    print('PUTVAL {}/exec-btrfs_{}/gauge-data_total {}:{:.1f}'.format(hostname, name, timestamp, size_data_total))
+    print('PUTVAL {}/exec-btrfs_{}/gauge-data_exclusive {}:{:.1f}'.format(hostname, name, timestamp, size_data_exclusive))
+    print('PUTVAL {}/exec-btrfs_{}/gauge-snapshot_total {}:{:.1f}'.format(hostname, name, timestamp, size_snapshot_total))
+    print('PUTVAL {}/exec-btrfs_{}/gauge-snapshot_exclusive {}:{:.1f}'.format(hostname, name, timestamp, size_snapshot_exclusive))
+
+
+#
+# Volumes to scan
+#
+hostname = 'server'
+interval = 10
+volumes = [
+    ['mezzo-scratch',   '/mnt/mezzo/scratch'],
+    ['mezzo-sync',      '/mnt/mezzo/sync'],
+    ['helium-personal', '/mnt/yotta/helium/personal'],
+    ['helium-shared',   '/mnt/yotta/helium/shared'],
+    ['neon',            '/mnt/yotta/neon'],
+    ['krypton',         '/mnt/yotta/krypton'],
+    ['xenon-borg',      '/mnt/yotta/xenon/borg'],
+    ['xenon-rsnapshot', '/mnt/yotta/xenon/rsnapshot']
+]
+
+
+#
+# Command line arguments
+#
+parser = argparse.ArgumentParser(description='Get BTRFS disk usage')
+parser.add_argument('-s', action='store_true', help='print in human readable format')
+args = parser.parse_args()
+human_readable = args.s
+
+
+#
+# Main
+#
+if human_readable:
+    for name, path in volumes:
+        get_disk_usage(name, path)
+        print_human_readable(name)
+else:
+    # RRD mode
+    while True:
+        for name, path in volumes:
+            get_disk_usage(name, path)        
+            print_rrd(name)
+
+        time.sleep(interval)
+        rescan_quota(path)