Add --verify option.

4 files changed, 41 insertions, 0 deletions

diff --git a/NEWS b/NEWS
index a864ba1..673bd7b 100644
--- a/NEWS
+++ b/NEWS
@@ -4,6 +4,10 @@ NEWS for seivot, a backup benchmarking program
 Version 1.13, released UNRELEASED
 ---------------------------------
 
+* `seivot` can now optionally verify that data is restored correctly:
+  see the `--verify` option. This is slow-ish, so it's not enabled
+  by default. You need the `summain` program installed for this
+  feature.
 * `seivots-summary` now has a manual page.
 
 Version 1.12, released 2011-08-24
diff --git a/debian/changelog b/debian/changelog
index 768bd16..fdf563c 100644
--- a/debian/changelog
+++ b/debian/changelog
@@ -1,3 +1,10 @@
+seivot (1.13-1) unstable; urgency=low
+
+  * New upstream release.
+  * Suggest summain, for the --verify option.
+
+ -- Lars Wirzenius <liw@liw.fi>  Thu, 25 Aug 2011 17:05:48 +0100
+
 seivot (1.12-1) unstable; urgency=low
 
   * New upstream release.
diff --git a/debian/control b/debian/control
index 2da51cb..35306a2 100644
--- a/debian/control
+++ b/debian/control
@@ -12,6 +12,7 @@ Package: seivot
 Architecture: all
 Depends: ${shlibs:Depends}, ${python:Depends}, ${misc:Depends},
     python-cliapp (>= 0.15)
+Suggests: summain
 Description: benchmark program for backup software
  seivot generates synthetic test data and backs it up using the
  desired backup tool. It then modifies the test data, and makes
diff --git a/seivot b/seivot
index 75c151d..14a7ca8 100755
--- a/seivot
+++ b/seivot
@@ -370,6 +370,8 @@ class Seivot(cliapp.Application):
                               'access backup repository over the '
                                 'network via sftp')
 
+        self.settings.boolean(['verify'], 'verifyt restored data')
+
     def process_args(self, args):
         progname = self.settings['program']
         logging.info('Benchmarking: %s' % progname)
@@ -406,11 +408,15 @@ class Seivot(cliapp.Application):
         else:
             amount = self.settings['initial-data']
             self.generate_live_data(self.live_data, amount)
+        if self.settings['verify']:
+            self.summain(self.live_data, 'backup-0.summain')
         self.measure(prog.backup, 0, amount)
 
         for i in range(1, generations):
             self.generate_live_data(self.live_data,
                                     self.settings['incremental-data'])
+            if self.settings['verify']:
+                self.summain(self.live_data, 'backup-%d.summain' % i)
             self.measure(prog.backup, i, self.settings['incremental-data'])
 
         for i in range(generations):
@@ -420,6 +426,7 @@ class Seivot(cliapp.Application):
             target_dir = os.path.join(self.tempdir, 'restored')
             os.mkdir(target_dir)
             self.measure(prog.restore, i, 0, target_dir=target_dir)
+            self.verify(target_dir, i)
             shutil.rmtree(target_dir)
 
         for i in range(generations):
@@ -436,6 +443,28 @@ class Seivot(cliapp.Application):
         runcmd(['genbackupdata', where, '--create', str(size),
                 '--file-size', str(self.settings['file-size'])])
 
+    def summain(self, dirname, basename):
+        '''Remember state of dirname at this time.
+        
+        This runs the summain(1) utility against dirname and stores
+        the result in a file called basename in the temporary directory.
+        
+        '''
+        
+        # We exclude mtime from summain output, because there are very
+        # small time differences that I choose to ignore at this time.
+        self.runcmd(['summain', '--relative', '--exclude=mtime', '--output',
+                     os.path.join(self.tempdir, basename), dirname])
+
+    def verify(self, dirname, generation):
+        '''Verify that the generation was restored correctly.'''
+
+        root = os.path.join(dirname, './' + self.live_data)
+        self.summain(root, 'restored-%d.summain' % generation)
+        orig = os.path.join(self.tempdir, 'backup-%d.summain' % generation)
+        rest = os.path.join(self.tempdir, 'restored-%d.summain' % generation)
+        self.runcmd(['diff', '-u', orig, rest])
+
     def file_sizes(self, dirname):
         bytes = 0
         for dirname, subdirs, basenames in os.walk(dirname):