Aurora Backup REPL

backup_repl.py

"""A REPL for performing ad-hoc analysis of Aurora storage backups."""
import code

from twitter.common import app

from gen.apache.aurora.storage.ttypes import Snapshot, TBinaryProtocol

from thrift.transport import TTransport


def load_backup(filename):
  with open(filename) as f:
    transport = TTransport.TFileObjectTransport(f)
    protocol = TBinaryProtocol.TBinaryProtocol(transport)
    backup = Snapshot()
    backup.read(protocol)
  return backup


def load_backups(filenames):
  return [load_backup(fname) for fname in filenames]


def main(args, _):
  backups = load_backups(args)
  code.interact('Backups REPL', local=locals())

app.set_name("backup_repl.pex")
app.set_usage("""backup_repl.pex [SNAPSHOT...]

Start a REPL for performing ad-hoc analysis of Aurora storage backups. Any files specified are
loaded into the ``backups'' list.

# Analyze a backup copied from /var/lib/mesos/scheduler_backups on a scheduler
% ./backup_repl.pex scheduler-backup-2013-05-22-19-03
Backups REPL

# Or equivalently
>>> backup = load_backup('scheduler-backup-2013-05-22-19-03')

# Number of stored tasks
>>> len(backups[0].tasks)
75632

# Number of stored jobs
>>> len(backups[0].jobs)
299

# Backup metadata
>>> backups[0].schedulerMetadata
SchedulerMetadata(frameworkId='201104070004-0000002563-0000')

# Arbitrary queries
>>> len([t for t in backups[0].tasks if t.assignedTask.task.jobName == 'hello_world'])
6
""")
app.main()