From: Brian Warner <warner@allmydata.com>
Date: Wed, 13 Aug 2008 00:05:01 +0000 (-0700)
Subject: CLI: add 'tahoe debug corrupt-share', and use it for deep-verify tests, and fix non... 
X-Git-Url: https://git.rkrishnan.org/components/architecture.txt?a=commitdiff_plain;h=014c9b59690fd54b9303be8289c6f1f5cbdc7b97;p=tahoe-lafs%2Ftahoe-lafs.git

CLI: add 'tahoe debug corrupt-share', and use it for deep-verify tests, and fix non-deep web checker API to pass verify=true into node
---

diff --git a/docs/CLI.txt b/docs/CLI.txt
index d2e6ede6..04a1a8b8 100644
--- a/docs/CLI.txt
+++ b/docs/CLI.txt
@@ -356,3 +356,7 @@ data for this file.
 Tahoe packages and modules are available on sys.path (e.g. by using 'import
 allmydata'). This is most useful from a source tree: it simply sets the
 PYTHONPATH correctly and runs the 'python' executable.
+
+"tahoe debug corrupt-share SHAREFILE" will flip a bit in the given sharefile.
+This can be used to test the client-side verification/repair code. Obviously
+this command should not be used during normal operation.
diff --git a/src/allmydata/scripts/debug.py b/src/allmydata/scripts/debug.py
index 4ca32aaa..243e49d8 100644
--- a/src/allmydata/scripts/debug.py
+++ b/src/allmydata/scripts/debug.py
@@ -619,6 +619,83 @@ def catalog_shares(options):
                         describe_share(abs_sharefile, si_s, shnum_s, now, out)
     return 0
 
+class CorruptShareOptions(usage.Options):
+    def getSynopsis(self):
+        return "Usage: tahoe debug corrupt-share SHARE_FILENAME"
+
+    optParameters = [
+        ["offset", "o", "block-random", "Which bit to flip."],
+        ]
+
+    def getUsage(self, width=None):
+        t = usage.Options.getUsage(self, width)
+        t += """
+Corrupt the given share by flipping a bit. This will cause a
+verifying/downloading client to log an integrity-check failure incident, and
+downloads will proceed with a different share.
+
+The --offset parameter controls which bit should be flipped. The default is
+to flip a single random bit of the block data.
+
+ tahoe debug corrupt-share testgrid/node-3/storage/shares/4v/4vozh77tsrw7mdhnj7qvp5ky74/0
+
+Obviously, this command should not be used in normal operation.
+"""
+        return t
+    def parseArgs(self, filename):
+        self['filename'] = filename
+
+def corrupt_share(options):
+    import random
+    from allmydata import storage
+    from allmydata.mutable.layout import unpack_header
+    out = options.stdout
+    fn = options['filename']
+    assert options["offset"] == "block-random", "other offsets not implemented"
+    # first, what kind of share is it?
+
+    def flip_bit(start, end):
+        offset = random.randrange(start, end)
+        bit = random.randrange(0, 8)
+        print >>out, "[%d..%d):  %d.b%d" % (start, end, offset, bit)
+        f = open(fn, "rb+")
+        f.seek(offset)
+        d = f.read(1)
+        d = chr(ord(d) ^ 0x01)
+        f.seek(offset)
+        f.write(d)
+        f.close()
+
+    f = open(fn, "rb")
+    prefix = f.read(32)
+    f.close()
+    if prefix == storage.MutableShareFile.MAGIC:
+        # mutable
+        m = storage.MutableShareFile(fn)
+        f = open(fn, "rb")
+        f.seek(m.DATA_OFFSET)
+        data = f.read(2000)
+        # make sure this slot contains an SMDF share
+        assert data[0] == "\x00", "non-SDMF mutable shares not supported"
+        f.close()
+
+        (version, ig_seqnum, ig_roothash, ig_IV, ig_k, ig_N, ig_segsize,
+         ig_datalen, offsets) = unpack_header(data)
+
+        assert version == 0, "we only handle v0 SDMF files"
+        start = m.DATA_OFFSET + offsets["share_data"]
+        end = m.DATA_OFFSET + offsets["enc_privkey"]
+        flip_bit(start, end)
+    else:
+        # otherwise assume it's immutable
+        f = storage.ShareFile(fn)
+        bp = storage.ReadBucketProxy(None)
+        offsets = bp._parse_offsets(f.read_share_data(0, 0x24))
+        start = f._data_offset + offsets["data"]
+        end = f._data_offset + offsets["plaintext_hash_tree"]
+        flip_bit(start, end)
+
+
 
 class ReplOptions(usage.Options):
     pass
@@ -635,6 +712,7 @@ class DebugCommand(usage.Options):
         ["dump-cap", None, DumpCapOptions, "Unpack a read-cap or write-cap"],
         ["find-shares", None, FindSharesOptions, "Locate sharefiles in node dirs"],
         ["catalog-shares", None, CatalogSharesOptions, "Describe shares in node dirs"],
+        ["corrupt-share", None, CorruptShareOptions, "Corrupt a share"],
         ["repl", None, ReplOptions, "Open a python interpreter"],
         ]
     def postOptions(self):
@@ -650,6 +728,7 @@ Subcommands:
     tahoe debug dump-cap        Unpack a read-cap or write-cap
     tahoe debug find-shares     Locate sharefiles in node directories
     tahoe debug catalog-shares  Describe all shares in node dirs
+    tahoe debug corrupt-share   Corrupt a share by flipping a bit.
 
 Please run e.g. 'tahoe debug dump-share --help' for more details on each
 subcommand.
@@ -661,6 +740,7 @@ subDispatch = {
     "dump-cap": dump_cap,
     "find-shares": find_shares,
     "catalog-shares": catalog_shares,
+    "corrupt-share": corrupt_share,
     "repl": repl,
     }
 
diff --git a/src/allmydata/test/test_system.py b/src/allmydata/test/test_system.py
index af64ec97..aa5846d3 100644
--- a/src/allmydata/test/test_system.py
+++ b/src/allmydata/test/test_system.py
@@ -1,5 +1,5 @@
 from base64 import b32encode
-import os, random, struct, sys, time, re, simplejson
+import os, random, struct, sys, time, re, simplejson, urllib
 from cStringIO import StringIO
 from twisted.trial import unittest
 from twisted.internet import defer
@@ -1851,3 +1851,71 @@ class ImmutableChecker(ShareManglingMixin, unittest.TestCase):
         d.addCallback(_check2)
         return d
     test_check_with_verify.todo = "We haven't implemented a verifier this thorough yet."
+
+class MutableChecker(SystemTestMixin, unittest.TestCase):
+
+    def _run_cli(self, argv):
+        stdout, stderr = StringIO(), StringIO()
+        runner.runner(argv, run_by_human=False, stdout=stdout, stderr=stderr)
+        return stdout.getvalue()
+
+    def test_good(self):
+        self.basedir = self.mktemp()
+        d = self.set_up_nodes()
+        CONTENTS = "a little bit of data"
+        d.addCallback(lambda res: self.clients[0].create_mutable_file(CONTENTS))
+        def _created(node):
+            self.node = node
+            si = self.node.get_storage_index()
+        d.addCallback(_created)
+        # now make sure the webapi verifier sees no problems
+        def _do_check(res):
+            url = (self.webish_url +
+                   "uri/%s" % urllib.quote(self.node.get_uri()) +
+                   "?t=check&verify=true")
+            return getPage(url, method="POST")
+        d.addCallback(_do_check)
+        def _got_results(out):
+            self.failUnless("<pre>Healthy!" in out, out)
+            self.failIf("Not Healthy!" in out, out)
+            self.failIf("Unhealthy" in out, out)
+            self.failIf("Corrupt Shares" in out, out)
+        d.addCallback(_got_results)
+        return d
+
+    def test_corrupt(self):
+        self.basedir = self.mktemp()
+        d = self.set_up_nodes()
+        CONTENTS = "a little bit of data"
+        d.addCallback(lambda res: self.clients[0].create_mutable_file(CONTENTS))
+        def _created(node):
+            self.node = node
+            si = self.node.get_storage_index()
+            out = self._run_cli(["debug", "find-shares", base32.b2a(si),
+                                self.clients[1].basedir])
+            files = out.split("\n")
+            # corrupt one of them, using the CLI debug command
+            f = files[0]
+            shnum = os.path.basename(f)
+            nodeid = self.clients[1].nodeid
+            nodeid_prefix = idlib.shortnodeid_b2a(nodeid)
+            self.corrupt_shareid = "%s-sh%s" % (nodeid_prefix, shnum)
+            out = self._run_cli(["debug", "corrupt-share", files[0]])
+        d.addCallback(_created)
+        # now make sure the webapi verifier notices it
+        def _do_check(res):
+            url = (self.webish_url +
+                   "uri/%s" % urllib.quote(self.node.get_uri()) +
+                   "?t=check&verify=true")
+            return getPage(url, method="POST")
+        d.addCallback(_do_check)
+        def _got_results(out):
+            self.failUnless("Not Healthy!" in out, out)
+            self.failUnless("Unhealthy: best recoverable version has only 9 shares (encoding is 3-of-10)" in out, out)
+            shid_re = (r"Corrupt Shares:\s+%s: block hash tree failure" %
+                       self.corrupt_shareid)
+            self.failUnless(re.search(shid_re, out), out)
+
+        d.addCallback(_got_results)
+        return d
+
diff --git a/src/allmydata/web/directory.py b/src/allmydata/web/directory.py
index 6adcd072..3b62b4a0 100644
--- a/src/allmydata/web/directory.py
+++ b/src/allmydata/web/directory.py
@@ -339,8 +339,7 @@ class DirectoryNodeHandler(RenderMixin, rend.Page, ReplaceMeMixin):
     def _POST_deep_check(self, req):
         # check this directory and everything reachable from it
         verify = boolean_of_arg(get_arg(req, "verify", "false"))
-        #repair = boolean_of_arg(get_arg(req, "repair", "false"))
-        repair = False # make sure it works first
+        repair = boolean_of_arg(get_arg(req, "repair", "false"))
         d = self.node.deep_check(verify, repair)
         d.addCallback(lambda res: DeepCheckResults(res))
         return d
diff --git a/src/allmydata/web/filenode.py b/src/allmydata/web/filenode.py
index 7996e757..185ab1dd 100644
--- a/src/allmydata/web/filenode.py
+++ b/src/allmydata/web/filenode.py
@@ -255,7 +255,9 @@ class FileNodeHandler(RenderMixin, rend.Page, ReplaceMeMixin):
         return d
 
     def _POST_check(self, req):
-        d = self.node.check()
+        verify = boolean_of_arg(get_arg(req, "verify", "false"))
+        repair = boolean_of_arg(get_arg(req, "repair", "false"))
+        d = self.node.check(verify, repair)
         d.addCallback(lambda res: CheckerResults(res))
         return d