Added external merging support

7 files changed, 196 insertions, 2 deletions

diff --git a/Makefile b/Makefile
index 78f8a6b..aa96290 100644
--- a/Makefile
+++ b/Makefile
@@ -10,8 +10,13 @@ release:
 		ln -s ../../plop/test .)
 	cp httpd_props.conf rel
 	cp catlfish.config rel
+	cp storage_node.config rel
+	cp storage_node_httpd.conf rel
 	mkdir rel/catlfish
 	mkdir rel/db
+	mkdir rel/mergedb
+	mkdir rel/mergedb/chains
+	touch rel/mergedb/logorder
 	printf "0" > rel/db/treesize
 	cp -r webroot rel/catlfish
 	test -d rel/catlfish/webroot/log || mkdir rel/catlfish/webroot/log
diff --git a/catlfish.config b/catlfish.config
index 75d00fa..672f997 100644
--- a/catlfish.config
+++ b/catlfish.config
@@ -21,4 +21,7 @@
    {index_path, "db/index"},
    {entryhash_root_path, "db/entryhash/"},
    {treesize_path, "db/treesize"},
-   {indexforhash_root_path, "db/certindex/"}]}].
+   {indexforhash_root_path, "db/certindex/"},
+   %{storage_nodes, ["https://127.0.0.1:8081/ct/storage/"]},
+   {storage_nodes_quorum, 1}
+  ]}].
diff --git a/httpd_props.conf b/httpd_props.conf
index ae4c7be..9ea7b30 100644
--- a/httpd_props.conf
+++ b/httpd_props.conf
@@ -8,7 +8,7 @@
  {document_root, "catlfish/webroot/docroot"},
  {modules, [mod_alias, mod_auth, mod_esi, mod_get, mod_head,
             mod_log, mod_disk_log]},
- {erl_script_alias, {"/ct", [v1]}},
+ {erl_script_alias, {"/ct", [v1, frontend]}},
  {erl_script_nocache, true},
  {error_log, "log/error"},
  {security_log, "log/security"},
diff --git a/storage_node.config b/storage_node.config
new file mode 100644
index 0000000..47a1326
--- /dev/null
+++ b/storage_node.config
@@ -0,0 +1,19 @@
+%% catlfish configuration file (-*- erlang -*-)
+%% Start like this:
+%% $ erl -boot start_sasl -config catlfish -run inets
+[{sasl,
+  [{sasl_error_logger, false},
+   {errlog_type, error},
+   {error_logger_mf_dir, "log"},
+   {error_logger_mf_maxbytes, 10485760},	% 10 MB
+   {error_logger_mf_maxfiles, 10}]},
+ {inets,
+  [{services,
+    [{httpd, [{proplist_file, "storage_node_httpd.conf"}]}]}]},
+ {plop,
+  [{entry_root_path, "db/certentries/"},
+   {index_path, "db/index"},
+   {newentries_path, "db/newentries"},
+   {entryhash_root_path, "db/entryhash/"},
+   {treesize_path, "db/treesize"},
+   {indexforhash_root_path, "db/certindex/"}]}].
diff --git a/storage_node_httpd.conf b/storage_node_httpd.conf
new file mode 100644
index 0000000..2f271f8
--- /dev/null
+++ b/storage_node_httpd.conf
@@ -0,0 +1,21 @@
+%%% Copyright (c) 2014, NORDUnet A/S.
+%%% See LICENSE for licensing information.
+[
+ {port, 8081},
+ {bind_address, {127, 0, 0, 1}},
+ {server_name, "flimsy"},
+ {server_root, "catlfish/webroot"},
+ {document_root, "catlfish/webroot/docroot"},
+ {modules, [mod_alias, mod_auth, mod_esi, mod_get, mod_head,
+            mod_log, mod_disk_log]},
+ {erl_script_alias, {"/ct", [storage]}},
+ {erl_script_nocache, true},
+ {error_log, "log/error_storage"},
+ {security_log, "log/security_storage"},
+ {transfer_log, "log/transfer_storage"},
+ {socket_type,
+  {essl,                              % See ssl(3erl) for SSL options.
+   [{certfile, "catlfish/webroot/certs/webcert.pem"},
+    {keyfile, "catlfish/webroot/keys/webkey.pem"},
+    {cacertfile, "catlfish/webroot/certs/webcert.pem"}]}}
+].
diff --git a/tools/merge.py b/tools/merge.py
new file mode 100755
index 0000000..7120d04
--- /dev/null
+++ b/tools/merge.py
@@ -0,0 +1,133 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+# Copyright (c) 2014 Kungliga Tekniska Högskolan
+# (KTH Royal Institute of Technology, Stockholm, Sweden).
+# See LICENSE for licensing information.
+
+import json
+import base64
+import urllib
+import urllib2
+import sys
+
+frontendnodes = ["https://127.0.0.1:8080/"]
+storagenodes = ["https://127.0.0.1:8081/"]
+
+chainsdir = "../rel/mergedb/chains"
+logorderfile = "../rel/mergedb/logorder"
+
+def parselogrow(row):
+    return base64.b16decode(row)
+
+def get_logorder():
+    f = open(logorderfile, "r")
+    return [parselogrow(row.rstrip()) for row in f]
+
+def write_chain(key, value):
+    f = open(chainsdir + "/" + base64.b16encode(key), "w")
+    f.write(value)
+    f.close()
+
+def read_chain(key):
+    f = open(chainsdir + "/" + base64.b16encode(key), "r")
+    value = f.read()
+    f.close()
+    return value
+
+def add_to_logorder(key):
+    f = open(logorderfile, "a")
+    f.write(base64.b16encode(key) + "\n")
+    f.close()
+
+def get_new_entries(baseurl):
+    try:
+        result = urllib2.urlopen(baseurl + "ct/storage/fetchnewentries").read()
+        parsed_result = json.loads(result)
+        if parsed_result.get(u"result") == u"ok":
+            return parsed_result[u"entries"]
+        print "ERROR: fetchnewentries", parsed_result
+        sys.exit(1)
+    except urllib2.HTTPError, e:
+        print "ERROR: fetchnewentries", e.read()
+        sys.exit(1)
+
+def get_curpos(baseurl):
+    try:
+        result = urllib2.urlopen(baseurl + "ct/frontend/currentposition").read()
+        parsed_result = json.loads(result)
+        if parsed_result.get(u"result") == u"ok":
+            return parsed_result[u"position"]
+        print "ERROR: currentposition", parsed_result
+        sys.exit(1)
+    except urllib2.HTTPError, e:
+        print "ERROR: currentposition", e.read()
+        sys.exit(1)
+
+def sendlog(baseurl, submission):
+    try:
+        result = urllib2.urlopen(baseurl + "ct/frontend/sendlog",
+            json.dumps(submission)).read()
+        return json.loads(result)
+    except urllib2.HTTPError, e:
+        print "ERROR: sendlog", e.read()
+        sys.exit(1)
+    except ValueError, e:
+        print "==== FAILED REQUEST ===="
+        print submission
+        print "======= RESPONSE ======="
+        print result
+        print "========================"
+        raise e
+
+def sendsth(baseurl, submission):
+    try:
+        result = urllib2.urlopen(baseurl + "ct/frontend/sendsth",
+            json.dumps(submission)).read()
+        return json.loads(result)
+    except urllib2.HTTPError, e:
+        print "ERROR: sendsth", e.read()
+        sys.exit(1)
+    except ValueError, e:
+        print "==== FAILED REQUEST ===="
+        print submission
+        print "======= RESPONSE ======="
+        print result
+        print "========================"
+        raise e
+
+def get_missingentries(baseurl):
+    try:
+        result = urllib2.urlopen(baseurl + "ct/frontend/missingentries").read()
+        parsed_result = json.loads(result)
+        if parsed_result.get(u"result") == u"ok":
+            return parsed_result[u"entries"]
+        print "ERROR: missingentries", parsed_result
+        sys.exit(1)
+    except urllib2.HTTPError, e:
+        print "ERROR: missingentries", e.read()
+        sys.exit(1)
+
+
+logorder = get_logorder()
+certsinlog = set(logorder)
+
+new_entries = [entry for storagenode in storagenodes for entry in get_new_entries(storagenode)]
+
+for new_entry in new_entries:
+    hash = base64.b64decode(new_entry["hash"])
+    entry = base64.b64decode(new_entry["entry"])
+    if hash not in certsinlog:
+        write_chain(hash, entry)
+        add_to_logorder(hash)
+        logorder.append(hash)
+        certsinlog.add(hash)
+        print "added", base64.b16encode(hash)
+
+for frontendnode in frontendnodes:
+    curpos = get_curpos(frontendnode)
+    entries = [base64.b64encode(entry) for entry in logorder[curpos:]]
+    sendlog(frontendnode, {"start": curpos, "hashes": entries})
+    missingentries = get_missingentries(frontendnode)
+    print "missing entries:", missingentries
+    # XXX: no test case for missing entries yet, waiting to implement
+    sendsth(frontendnode, {"tree_size": len(logorder)})
diff --git a/tools/testcase1.py b/tools/testcase1.py
index eab6c6f..2d5e0e8 100755
--- a/tools/testcase1.py
+++ b/tools/testcase1.py
@@ -1,4 +1,5 @@
 #!/usr/bin/env python
+# -*- coding: utf-8 -*-
 
 # Copyright (c) 2014, NORDUnet A/S.
 # See LICENSE for licensing information.
@@ -125,12 +126,16 @@ testgroup("cert1")
 
 result1 = do_add_chain(cc1)
 
+subprocess.call(["./merge.py"])
+
 print_and_check_tree_size(1)
 
 result2 = do_add_chain(cc1)
 
 assert_equal(result2["timestamp"], result1["timestamp"], "timestamp")
 
+subprocess.call(["./merge.py"])
+
 print_and_check_tree_size(1)
 
 # TODO: add invalid cert and check that it generates an error
@@ -142,6 +147,8 @@ testgroup("cert2")
 
 result3 = do_add_chain(cc2)
 
+subprocess.call(["./merge.py"])
+
 print_and_check_tree_size(2)
 
 get_and_validate_proof(result1["timestamp"], cc1, 0, 1)
@@ -151,6 +158,8 @@ testgroup("cert3")
 
 result4 = do_add_chain(cc3)
 
+subprocess.call(["./merge.py"])
+
 print_and_check_tree_size(3)
 
 get_and_validate_proof(result1["timestamp"], cc1, 0, 2)
@@ -161,6 +170,8 @@ testgroup("cert4")
 
 result5 = do_add_chain(cc4)
 
+subprocess.call(["./merge.py"])
+
 print_and_check_tree_size(4)
 
 get_and_validate_proof(result1["timestamp"], cc1, 0, 2)
@@ -172,6 +183,8 @@ testgroup("cert5")
 
 result6 = do_add_chain(cc5)
 
+subprocess.call(["./merge.py"])
+
 print_and_check_tree_size(5)
 
 get_and_validate_proof(result1["timestamp"], cc1, 0, 3)