4342601d00db2d8d75a020fd9684ad9bf7424a4e
[cumulus.git] / lbs-util
1 #!/usr/bin/python
2 #
3 # Utility for managing LBS archives.
4
5 import getpass, os, sys
6 from optparse import OptionParser
7 import lbs
8
9 parser = OptionParser(usage="%prog [option]... command [arg]...")
10 parser.add_option("-v", action="store_true", dest="verbose", default=False,
11                   help="increase verbosity")
12 parser.add_option("--store", dest="store",
13                   help="specify path to backup data store")
14 parser.add_option("--localdb", dest="localdb",
15                   help="specify path to local database")
16 (options, args) = parser.parse_args(sys.argv[1:])
17
18 # Read a passphrase from the user and store it in the LBS_GPG_PASSPHRASE
19 # environment variable.
20 def get_passphrase():
21     ENV_KEY = 'LBS_GPG_PASSPHRASE'
22     if not os.environ.has_key(ENV_KEY):
23         os.environ[ENV_KEY] = getpass.getpass()
24
25 # Delete old snapshots from the local database, though do not actually schedule
26 # any segment cleaning.
27 # Syntax: $0 --localdb=LOCALDB prune-db
28 def cmd_prune_db():
29     db = lbs.LocalDatabase(options.localdb)
30
31     # Delete old snapshots from the local database.
32     db.garbage_collect()
33     db.commit()
34
35 # Run the segment cleaner.
36 # Syntax: $0 --localdb=LOCALDB clean
37 def cmd_clean(clean_threshold=7.0):
38     db = lbs.LocalDatabase(options.localdb)
39
40     # Delete old snapshots from the local database.
41     db.garbage_collect()
42
43     # Expire segments which are poorly-utilized.
44     for s in db.get_segment_cleaning_list():
45         if s.cleaning_benefit > clean_threshold:
46             print "Cleaning segment %d (benefit %.2f)" % (s.id,
47                                                           s.cleaning_benefit)
48             db.mark_segment_expired(s)
49         else:
50             break
51     db.balance_expired_objects()
52     db.commit()
53
54 # List snapshots stored.
55 # Syntax: $0 --data=DATADIR list-snapshots
56 def cmd_list_snapshots():
57     store = lbs.LowlevelDataStore(options.store)
58     for s in sorted(store.list_snapshots()):
59         print s
60
61 # List size of data needed for each snapshot.
62 # Syntax: $0 --data=DATADIR list-snapshot-sizes
63 def cmd_list_snapshot_sizes():
64     lowlevel = lbs.LowlevelDataStore(options.store)
65     store = lbs.ObjectStore(lowlevel)
66     previous = set()
67     for s in sorted(lowlevel.list_snapshots()):
68         d = lbs.parse_full(store.load_snapshot(s))
69         segments = d['Segments'].split()
70         (size, added, removed) = (0, 0, 0)
71         for seg in segments:
72             segsize = lowlevel.lowlevel_stat(seg + ".tar.gpg")['size']
73             size += segsize
74             if seg not in previous: added += segsize
75         for seg in previous:
76             if seg not in segments:
77                 removed += lowlevel.lowlevel_stat(seg + ".tar.gpg")['size']
78         previous = set(segments)
79         print "%s: %.3f +%.3f -%.3f" % (s, size / 1024.0**2, added / 1024.0**2, removed / 1024.0**2)
80
81 # Build checksum list for objects in the given segments, or all segments if
82 # none are specified.
83 def cmd_object_checksums(segments):
84     get_passphrase()
85     lowlevel = lbs.LowlevelDataStore(options.store)
86     store = lbs.ObjectStore(lowlevel)
87     if len(segments) == 0:
88         segments = sorted(lowlevel.list_segments())
89     for s in segments:
90         for (o, data) in store.load_segment(s):
91             csum = lbs.ChecksumCreator().update(data).compute()
92             print "%s/%s:%d:%s" % (s, o, len(data), csum)
93     store.cleanup()
94
95 # Read a snapshot file
96 def cmd_read_snapshots(snapshots):
97     get_passphrase()
98     lowlevel = lbs.LowlevelDataStore(options.store)
99     store = lbs.ObjectStore(lowlevel)
100     for s in snapshots:
101         d = lbs.parse_full(store.load_snapshot(s))
102         print d
103         print d['Segments'].split()
104     store.cleanup()
105
106 # Verify snapshot integrity
107 def cmd_verify_snapshots(snapshots):
108     get_passphrase()
109     lowlevel = lbs.LowlevelDataStore(options.store)
110     store = lbs.ObjectStore(lowlevel)
111     for s in snapshots:
112         print "#### Snapshot", s
113         d = lbs.parse_full(store.load_snapshot(s))
114         print "## Root:", d['Root']
115         metadata = lbs.iterate_metadata(store, d['Root'])
116         for m in metadata:
117             if m.fields['type'] != '-': continue
118             print "%s [%d bytes]" % (m.fields['name'], int(m.fields['size']))
119             verifier = lbs.ChecksumVerifier(m.fields['checksum'])
120             size = 0
121             for block in m.data():
122                 data = store.get(block)
123                 verifier.update(data)
124                 size += len(data)
125             if int(m.fields['size']) != size:
126                 raise ValueError("File size does not match!")
127             if not verifier.valid():
128                 raise ValueError("Bad checksum found")
129     store.cleanup()
130
131 if len(args) == 0:
132     parser.print_usage()
133     sys.exit(1)
134 cmd = args[0]
135 args = args[1:]
136 if cmd == 'clean':
137     cmd_clean()
138 elif cmd == 'prune-db':
139     cmd_prune_db()
140 elif cmd == 'list-snapshots':
141     cmd_list_snapshots()
142 elif cmd == 'object-sums':
143     cmd_object_checksums(args)
144 elif cmd == 'read-snapshots':
145     cmd_read_snapshots(args)
146 elif cmd == 'list-snapshot-sizes':
147     cmd_list_snapshot_sizes()
148 elif cmd == 'verify-snapshots':
149     cmd_verify_snapshots(args)
150 else:
151     print "Unknown command:", cmd
152     parser.print_usage()
153     sys.exit(1)