Merge is now run by shell script tools/merge.

tools/merge run merge_fetch.py, merge_backup.py, merge_sth.py and
merge_dist.py sequentially.

TODO: test backupquorum != 0

9 files changed, 331 insertions, 115 deletions

diff --git a/Makefile b/Makefile
index 2f823e5..92d4577 100644
--- a/Makefile
+++ b/Makefile
@@ -76,7 +76,7 @@ tests-start:
 
 tests-run:
 	@(cd $(INSTDIR) && python ../tools/testcase1.py https://localhost:8080/ tests/keys/logkey.pem tests/httpsca/demoCA/cacert.pem) || (echo "Tests failed" ; false)
-	@diff -r -x nursery -x verifiedsize catlfish/tests/mergedb catlfish/tests/mergedb-secondary || (echo "Merge databases not matching" ; false)
+	@diff -r -x nursery -x verifiedsize -x fetched -x sth -x verified.\* catlfish/tests/mergedb catlfish/tests/mergedb-secondary || (echo "Merge databases not matching" ; false)
 	@(cd $(INSTDIR) && python ../tools/check-sth.py --publickey=tests/keys/logkey.pem --cafile tests/httpsca/demoCA/cacert.pem https://localhost:8080/) || (echo "Check failed" ; false)
 	@(cd $(INSTDIR) && python ../tools/fetchallcerts.py $(BASEURL) --publickey=tests/keys/logkey.pem --cafile tests/httpsca/demoCA/cacert.pem) || (echo "Verification failed" ; false)
 	@(cd $(INSTDIR) && rm -f submittedcerts)
@@ -89,8 +89,8 @@ tests-run:
 	@(cd $(INSTDIR) && python ../tools/submitcert.py --parallel=1 --store ../tools/testcerts/pre1.txt --check-sct --sct-file=submittedcerts $(BASEURL) --publickey=tests/keys/logkey.pem --cafile tests/httpsca/demoCA/cacert.pem) || (echo "Submission failed" ; false)
 	@(cd $(INSTDIR) && python ../tools/submitcert.py --parallel=1 --store ../tools/testcerts/pre2.txt --check-sct --sct-file=submittedcerts $(BASEURL) --publickey=tests/keys/logkey.pem --cafile tests/httpsca/demoCA/cacert.pem) || (echo "Submission failed" ; false)
 	@(cd $(INSTDIR) && python ../tools/storagegc.py --config ../test/catlfish-test.cfg --localconfig ../test/catlfish-test-local-1.cfg)  || (echo "GC failed" ; false)
-	@(cd $(INSTDIR) && python ../tools/merge.py --config ../test/catlfish-test.cfg --localconfig ../test/catlfish-test-local-merge.cfg) || (echo "Merge failed" ; false)
-	@diff -r -x nursery -x verifiedsize catlfish/tests/mergedb catlfish/tests/mergedb-secondary || (echo "Merge databases not matching" ; false)
+	@(cd $(INSTDIR) && ../tools/merge --config ../test/catlfish-test.cfg --localconfig ../test/catlfish-test-local-merge.cfg) || (echo "Merge failed" ; false)
+	@diff -r -x nursery -x verifiedsize -x fetched -x sth -x verified.\* catlfish/tests/mergedb catlfish/tests/mergedb-secondary || (echo "Merge databases not matching" ; false)
 	@(cd $(INSTDIR) && python ../tools/check-sth.py --publickey=tests/keys/logkey.pem --cafile tests/httpsca/demoCA/cacert.pem https://localhost:8080/) || (echo "Check failed" ; false)
 	@(cd $(INSTDIR) && rm -r fetchcertstore || true)
 	@(cd $(INSTDIR) && mkdir fetchcertstore)
@@ -148,8 +148,8 @@ tests-formatupgrade-prepare2:
 	printf 0 > $(INSTDIR)/tests/mergedb-secondary/verifiedsize
 
 tests-formatupgrade-run2:
-	@(cd $(INSTDIR) && python ../tools/merge.py --config ../test/catlfish-test.cfg --localconfig ../test/catlfish-test-local-merge.cfg) || (echo "Merge failed" ; false)
-	@diff -r -x nursery -x verifiedsize catlfish/tests/mergedb catlfish/tests/mergedb-secondary || (echo "Merge databases not matching" ; false)
+	@(cd $(INSTDIR) && ../tools/merge --config ../test/catlfish-test.cfg --localconfig ../test/catlfish-test-local-merge.cfg) || (echo "Merge failed" ; false)
+	@diff -r -x nursery -x verifiedsize -x fetched -x sth -x verified.\* catlfish/tests/mergedb catlfish/tests/mergedb-secondary || (echo "Merge databases not matching" ; false)
 	@(cd $(INSTDIR) && rm -r fetchcertstore-old || true)
 	@(cd $(INSTDIR) && mkdir fetchcertstore-old)
 	@(cd $(INSTDIR) && python ../tools/fetchallcerts.py $(BASEURL) --no-check-signature --store fetchcertstore-old --publickey=tests/keys/logkey.pem --cafile tests/httpsca/demoCA/cacert.pem) || (echo "Verification failed" ; false)
diff --git a/tools/merge b/tools/merge
new file mode 100755
index 0000000..e11ac93
--- /dev/null
+++ b/tools/merge
@@ -0,0 +1,8 @@
+#! /bin/sh
+
+BINDIR=$(dirname $0)
+
+$BINDIR/merge_fetch.py $@
+$BINDIR/merge_backup.py $@
+$BINDIR/merge_sth.py $@
+$BINDIR/merge_dist.py $@
diff --git a/tools/merge.py b/tools/merge.py
deleted file mode 100755
index 212c171..0000000
--- a/tools/merge.py
+++ /dev/null
@@ -1,38 +0,0 @@
-#!/usr/bin/env python
-# -*- coding: utf-8 -*-
-#
-# Copyright (c) 2014-2015, NORDUnet A/S.
-# See LICENSE for licensing information.
-
-import argparse
-import yaml
-import sys
-from certtools import create_ssl_context
-from merge_fetch import merge_fetch
-from merge_backup import merge_backup
-from merge_dist import merge_dist
-
-def main():
-    parser = argparse.ArgumentParser(description="")
-    parser.add_argument('--config', help="System configuration",
-                        required=True)
-    parser.add_argument('--localconfig', help="Local configuration",
-                        required=True)
-    parser.add_argument("--nomerge", action='store_true',
-                        help="Don't actually do merge")
-    parser.add_argument("--timing", action='store_true',
-                        help="Print timing information")
-    args = parser.parse_args()
-
-    config = yaml.load(open(args.config))
-    localconfig = yaml.load(open(args.localconfig))
-    paths = localconfig["paths"]
-
-    create_ssl_context(cafile=paths["https_cacertfile"])
-
-    sth = merge_fetch(args, config, localconfig)
-    merge_backup(args, config, localconfig, sth)
-    merge_dist(args, config, localconfig, sth)
-
-if __name__ == '__main__':
-    sys.exit(main())
diff --git a/tools/merge_backup.py b/tools/merge_backup.py
index 27c71a5..48197fc 100644..100755
--- a/tools/merge_backup.py
+++ b/tools/merge_backup.py
@@ -7,12 +7,15 @@
 import sys
 import base64
 import select
-from certtools import timing_point
+from time import sleep
+from certtools import timing_point, build_merkle_tree, write_file, \
+     create_ssl_context
 from mergetools import chunks, backup_sendlog, get_logorder, \
      get_verifiedsize, get_missingentriesforbackup, read_chain, \
-     hexencode, setverifiedsize, sendentry_merge, verifyroot
+     hexencode, setverifiedsize, sendentry_merge, verifyroot, \
+     get_nfetched, parse_args
 
-def merge_backup(args, config, localconfig, sth_in):
+def merge_backup(args, config, localconfig):
     paths = localconfig["paths"]
     own_key = (localconfig["nodename"],
                "%s/%s-private.pem" % (paths["privatekeys"],
@@ -21,12 +24,17 @@ def merge_backup(args, config, localconfig, sth_in):
     mergedb = paths["mergedb"]
     chainsdir = mergedb + "/chains"
     logorderfile = mergedb + "/logorder"
+    currentsizefile = mergedb + "/fetched"
     timing = timing_point()
 
-    logorder = get_logorder(logorderfile)
+    nfetched = get_nfetched(currentsizefile, logorderfile)
+    logorder = get_logorder(logorderfile, nfetched)
+    tree_size = len(logorder)
     timing_point(timing, "get logorder")
 
-    (tree_size, root_hash, _) = sth_in
+    tree = build_merkle_tree(logorder)
+    root_hash = tree[-1][0]
+    timing_point(timing, "build tree")
 
     for secondary in secondaries:
         if secondary["name"] == config["primarymergenode"]:
@@ -40,6 +48,7 @@ def merge_backup(args, config, localconfig, sth_in):
         timing_point(timing, "get verified size")
         print >>sys.stderr, "verified size", verifiedsize
         sys.stderr.flush()
+
         entries = [base64.b64encode(entry) for entry in logorder[verifiedsize:]]
         print >>sys.stderr, "sending log:",
         sys.stderr.flush()
@@ -57,7 +66,7 @@ def merge_backup(args, config, localconfig, sth_in):
                     continue
                 break
             if sendlogresult["result"] != "ok":
-                print >>sys.stderr, "sendlog:", sendlogresult
+                print >>sys.stderr, "backup_sendlog:", sendlogresult
                 sys.exit(1)
             verifiedsize += len(chunk)
             print >>sys.stderr, verifiedsize,
@@ -66,11 +75,13 @@ def merge_backup(args, config, localconfig, sth_in):
         timing_point(timing, "sendlog")
         print >>sys.stderr, "log sent"
         sys.stderr.flush()
+
         missingentries = get_missingentriesforbackup(nodename, nodeaddress,
                                                      own_key, paths)
         timing_point(timing, "get missing")
         print >>sys.stderr, "missing entries:", len(missingentries)
         sys.stderr.flush()
+
         fetched_entries = 0
         print >>sys.stderr, "fetching missing entries",
         sys.stderr.flush()
@@ -81,7 +92,7 @@ def merge_backup(args, config, localconfig, sth_in):
                                               read_chain(chainsdir, ehash),
                                               ehash)
             if sendentryresult["result"] != "ok":
-                print >>sys.stderr, "send sth:", sendentryresult
+                print >>sys.stderr, "sendentry_merge:", sendentryresult
                 sys.exit(1)
             fetched_entries += 1
             if fetched_entries % 1000 == 0:
@@ -90,6 +101,7 @@ def merge_backup(args, config, localconfig, sth_in):
         print >>sys.stderr
         sys.stderr.flush()
         timing_point(timing, "send missing")
+
         verifyrootresult = verifyroot(nodename, nodeaddress, own_key, paths,
                                       tree_size)
         if verifyrootresult["result"] != "ok":
@@ -102,7 +114,39 @@ def merge_backup(args, config, localconfig, sth_in):
             print >>sys.stderr, "  expected", hexencode(root_hash)
             sys.exit(1)
         timing_point(timing, "verifyroot")
+
         setverifiedsize(nodename, nodeaddress, own_key, paths, tree_size)
+        backuppath = mergedb + "/verified." + nodename
+        backupdata = {"tree_size": tree_size,
+                      "sha256_root_hash": hexencode(root_hash)}
+        print >>sys.stderr, "DEBUG: writing to", backuppath, ":", backupdata
+        write_file(backuppath, backupdata)
+
         if args.timing:
             print >>sys.stderr, timing["deltatimes"]
             sys.stderr.flush()
+
+def main():
+    """
+    Read logorder file up until what's indicated by fetched file and
+    build the tree.
+
+    Distribute entries to all secondaries, write tree size and tree head
+    to backup.<secondary> files as each secondary is verified to have
+    the entries.
+
+    Sleep some and start over.
+    """
+    args, config, localconfig = parse_args()
+    paths = localconfig["paths"]
+    create_ssl_context(cafile=paths["https_cacertfile"])
+
+    while True:
+        merge_backup(args, config, localconfig)
+        if args.interval is None:
+            break
+        print >>sys.stderr, "sleeping", args.interval, "seconds"
+        sleep(args.interval)
+
+if __name__ == '__main__':
+    sys.exit(main())
diff --git a/tools/merge_dist.py b/tools/merge_dist.py
index bfc0e61..0e85984 100644..100755
--- a/tools/merge_dist.py
+++ b/tools/merge_dist.py
@@ -5,74 +5,56 @@
 # See LICENSE for licensing information.
 
 import sys
-import urllib2
-import base64
-import select
-from certtools import timing_point, check_sth_signature, \
-     create_sth_signature, get_public_key_from_file
+import json
+from time import sleep
+from base64 import b64encode, b64decode
+from certtools import timing_point, \
+     create_ssl_context
 from mergetools import get_curpos, get_logorder, chunks, get_missingentries, \
-     sendsth, hexencode, sendlog, sendentry, read_chain
+     sendsth, sendlog, sendentry, read_chain, parse_args
 
-def merge_dist(args, config, localconfig, sth_in):
+def merge_dist(args, config, localconfig, timestamp):
     paths = localconfig["paths"]
     own_key = (localconfig["nodename"],
                "%s/%s-private.pem" % (paths["privatekeys"],
                                       localconfig["nodename"]))
     frontendnodes = config["frontendnodes"]
-    signingnodes = config["signingnodes"]
-    ctbaseurl = config["baseurl"]
-    logpublickey = get_public_key_from_file(paths["logpublickey"])
     mergedb = paths["mergedb"]
     chainsdir = mergedb + "/chains"
     logorderfile = mergedb + "/logorder"
+    sthfile = mergedb + "/sth"
+    create_ssl_context(cafile=paths["https_cacertfile"])
     timing = timing_point()
 
-    logorder = get_logorder(logorderfile)
-    timing_point(timing, "get logorder")
-
-    (tree_size, root_hash, timestamp) = sth_in
-    tree_head_signature = None
-    for signingnode in signingnodes:
-        try:
-            tree_head_signature = \
-              create_sth_signature(tree_size, timestamp,
-                                   root_hash,
-                                   "https://%s/" % signingnode["address"],
-                                   key=own_key)
-            break
-        except urllib2.URLError, err:
-            print >>sys.stderr, err
-            sys.stderr.flush()
-    if tree_head_signature == None:
-        print >>sys.stderr, "Could not contact any signing nodes"
-        sys.exit(1)
-
-    sth = {"tree_size": tree_size, "timestamp": timestamp,
-           "sha256_root_hash": base64.b64encode(root_hash),
-           "tree_head_signature": base64.b64encode(tree_head_signature)}
-
-    check_sth_signature(ctbaseurl, sth, publickey=logpublickey)
+    try:
+        sth = json.loads(open(sthfile, 'r').read())
+    except (IOError, ValueError):
+        print >>sys.stderr, "No valid STH file found in", sthfile
+        return timestamp
+    if sth['timestamp'] < timestamp:
+        print >>sys.stderr, "New STH file older than the previous one:", \
+          sth['timestamp'], "<", timestamp
+        return timestamp
+    if sth['timestamp'] == timestamp:
+        return timestamp
+    timestamp = sth['timestamp']
 
-    timing_point(timing, "build sth")
-
-    if args.timing:
-        print >>sys.stderr, timing["deltatimes"]
-        sys.stderr.flush()
-
-    print hexencode(root_hash)
-    sys.stdout.flush()
+    logorder = get_logorder(logorderfile, sth['tree_size'])
+    timing_point(timing, "get logorder")
 
     for frontendnode in frontendnodes:
         nodeaddress = "https://%s/" % frontendnode["address"]
         nodename = frontendnode["name"]
         timing = timing_point()
+
         print >>sys.stderr, "distributing for node", nodename
         sys.stderr.flush()
         curpos = get_curpos(nodename, nodeaddress, own_key, paths)
         timing_point(timing, "get curpos")
         print >>sys.stderr, "current position", curpos
         sys.stderr.flush()
-        entries = [base64.b64encode(entry) for entry in logorder[curpos:]]
+
+        entries = [b64encode(entry) for entry in logorder[curpos:]]
         print >>sys.stderr, "sending log:",
         sys.stderr.flush()
         for chunk in chunks(entries, 1000):
@@ -83,7 +65,7 @@ def merge_dist(args, config, localconfig, sth_in):
                 if sendlogresult == None:
                     if trynumber == 1:
                         sys.exit(1)
-                    select.select([], [], [], 10.0)
+                    sleep(10)
                     print >>sys.stderr, "tries left:", trynumber
                     sys.stderr.flush()
                     continue
@@ -98,20 +80,22 @@ def merge_dist(args, config, localconfig, sth_in):
         timing_point(timing, "sendlog")
         print >>sys.stderr, "log sent"
         sys.stderr.flush()
+
         missingentries = get_missingentries(nodename, nodeaddress, own_key,
                                             paths)
         timing_point(timing, "get missing")
+
         print >>sys.stderr, "missing entries:", len(missingentries)
         sys.stderr.flush()
         fetched_entries = 0
         print >>sys.stderr, "fetching missing entries",
         sys.stderr.flush()
         for missingentry in missingentries:
-            ehash = base64.b64decode(missingentry)
+            ehash = b64decode(missingentry)
             sendentryresult = sendentry(nodename, nodeaddress, own_key, paths,
                                         read_chain(chainsdir, ehash), ehash)
             if sendentryresult["result"] != "ok":
-                print >>sys.stderr, "send sth:", sendentryresult
+                print >>sys.stderr, "sendentry:", sendentryresult
                 sys.exit(1)
             fetched_entries += 1
             if fetched_entries % 1000 == 0:
@@ -120,11 +104,33 @@ def merge_dist(args, config, localconfig, sth_in):
         print >>sys.stderr
         sys.stderr.flush()
         timing_point(timing, "send missing")
+
+        print >>sys.stderr, "sending sth to node", nodename
+        sys.stderr.flush()
         sendsthresult = sendsth(nodename, nodeaddress, own_key, paths, sth)
         if sendsthresult["result"] != "ok":
-            print >>sys.stderr, "send sth:", sendsthresult
+            print >>sys.stderr, "sendsth:", sendsthresult
             sys.exit(1)
         timing_point(timing, "send sth")
+
         if args.timing:
             print >>sys.stderr, timing["deltatimes"]
             sys.stderr.flush()
+
+    return timestamp
+
+def main():
+    """
+    Distribute missing entries and the STH to all frontend nodes.
+    """
+    args, config, localconfig = parse_args()
+    timestamp = 0
+    while True:
+        timestamp = merge_dist(args, config, localconfig, timestamp)
+        if args.interval is None:
+            break
+        print >>sys.stderr, "sleeping", args.interval, "seconds"
+        sleep(args.interval)
+
+if __name__ == '__main__':
+    sys.exit(main())
diff --git a/tools/merge_fetch.py b/tools/merge_fetch.py
index a0a0396..1540c34 100644..100755
--- a/tools/merge_fetch.py
+++ b/tools/merge_fetch.py
@@ -6,13 +6,14 @@
 
 import sys
 import struct
-import time
 import subprocess
+from time import sleep
 from mergetools import get_logorder, verify_entry, get_new_entries, \
-     chunks, fsync_logorder, get_entries, write_chain, add_to_logorder
-from certtools import timing_point, build_merkle_tree
+     chunks, fsync_logorder, get_entries, write_chain, add_to_logorder, \
+     hexencode, parse_args
+from certtools import timing_point, write_file, create_ssl_context
 
-def merge_fetch(args, config, localconfig):
+def merge_fetch(_args, config, localconfig):
     paths = localconfig["paths"]
     storagenodes = config["storagenodes"]
     mergedb = paths["mergedb"]
@@ -21,7 +22,6 @@ def merge_fetch(args, config, localconfig):
     own_key = (localconfig["nodename"],
                "%s/%s-private.pem" % (paths["privatekeys"],
                                       localconfig["nodename"]))
-
     timing = timing_point()
 
     logorder = get_logorder(logorderfile)
@@ -48,9 +48,6 @@ def merge_fetch(args, config, localconfig):
     print >>sys.stderr, "adding", len(new_entries), "entries"
     sys.stderr.flush()
 
-    if args.nomerge:
-        sys.exit(0)
-
     for ehash in new_entries:
         for storagenode in storagenodes:
             if ehash in new_entries_per_node[storagenode["name"]]:
@@ -89,9 +86,36 @@ def merge_fetch(args, config, localconfig):
 
     verifycert.communicate(struct.pack("I", 0))
 
-    tree = build_merkle_tree(logorder)
     tree_size = len(logorder)
-    root_hash = tree[-1][0]
-    timestamp = int(time.time() * 1000)
+    if tree_size == 0:
+        return (0, '')
+    else:
+        return (tree_size, logorder[tree_size-1])
+
+def main():
+    """
+    Fetch new entries from all storage nodes.
+
+    Indicate current position by writing the index in the logorder file
+    (0-based) to the 'fetched' file.
+
+    Sleep some and start over.
+    """
+    args, config, localconfig = parse_args()
+    paths = localconfig["paths"]
+    mergedb = paths["mergedb"]
+    currentsizefile = mergedb + "/fetched"
+    create_ssl_context(cafile=paths["https_cacertfile"])
+
+    while True:
+        logsize, last_hash = merge_fetch(args, config, localconfig)
+        currentsize = {"index": logsize - 1, "hash": hexencode(last_hash)}
+        print >>sys.stderr, "DEBUG: writing to", currentsizefile, ":", currentsize
+        write_file(currentsizefile, currentsize)
+        if args.interval is None:
+            break
+        print >>sys.stderr, "sleeping", args.interval, "seconds"
+        sleep(args.interval)
 
-    return (tree_size, root_hash, timestamp)
+if __name__ == '__main__':
+    sys.exit(main())
diff --git a/tools/merge_sth.py b/tools/merge_sth.py
new file mode 100755
index 0000000..68b52a0
--- /dev/null
+++ b/tools/merge_sth.py
@@ -0,0 +1,123 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+#
+# Copyright (c) 2014-2015, NORDUnet A/S.
+# See LICENSE for licensing information.
+
+import sys
+import json
+import urllib2
+import time
+from base64 import b64encode, b64decode
+from mergetools import parse_args, get_nfetched, hexencode, hexdecode, \
+     get_logorder, get_sth
+from certtools import create_ssl_context, get_public_key_from_file, \
+     timing_point, create_sth_signature, write_file, check_sth_signature, \
+     build_merkle_tree
+
+def merge_sth(args, config, localconfig):
+    paths = localconfig["paths"]
+    own_key = (localconfig["nodename"],
+               "%s/%s-private.pem" % (paths["privatekeys"],
+                                      localconfig["nodename"]))
+    ctbaseurl = config["baseurl"]
+    signingnodes = config["signingnodes"]
+    mergenodes = config.get("mergenodes", [])
+    mergedb = paths["mergedb"]
+    sthfile = mergedb + "/sth"
+    logorderfile = mergedb + "/logorder"
+    logpublickey = get_public_key_from_file(paths["logpublickey"])
+    backupquorum = localconfig.get("backupquorum", 0)
+    assert backupquorum <= len(mergenodes) - 1
+    create_ssl_context(cafile=paths["https_cacertfile"])
+    timing = timing_point()
+
+    trees = [{'tree_size': 0, 'sha256_root_hash': ''}]
+    for mergenode in mergenodes:
+        if mergenode["name"] == config["primarymergenode"]:
+            continue
+        verifiedfile = mergedb + "/verified." + mergenode["name"]
+        try:
+            tree = json.loads(open(verifiedfile, "r").read())
+        except (IOError, ValueError):
+            tree = {'tree_size': 0, "sha256_root_hash": ''}
+        trees.append(tree)
+    trees.sort(key=lambda e: e['tree_size'], reverse=True)
+    print >>sys.stderr, "DEBUG: trees:", trees
+    tree_size = trees[backupquorum]['tree_size']
+    root_hash = hexdecode(trees[backupquorum]['sha256_root_hash'])
+    print >>sys.stderr, "DEBUG: tree size candidate at backupquorum", \
+      backupquorum, ":", tree_size
+
+    cur_sth = get_sth(sthfile)
+    if tree_size < cur_sth['tree_size']:
+        print >>sys.stderr, "candidate tree < current tree:", \
+          tree_size, "<", cur_sth['tree_size']
+        return
+
+    assert tree_size >= 0         # Don't read logorder without limit.
+    logorder = get_logorder(logorderfile, tree_size)
+    timing_point(timing, "get logorder")
+    if tree_size == -1:
+        tree_size = len(logorder)
+    print >>sys.stderr, "new tree size will be", tree_size
+
+    root_hash_calc = build_merkle_tree(logorder)[-1][0]
+    assert root_hash == '' or root_hash == root_hash_calc
+    root_hash = root_hash_calc
+    timestamp = int(time.time() * 1000)
+
+    tree_head_signature = None
+    for signingnode in signingnodes:
+        try:
+            tree_head_signature = \
+              create_sth_signature(tree_size, timestamp,
+                                   root_hash,
+                                   "https://%s/" % signingnode["address"],
+                                   key=own_key)
+            break
+        except urllib2.URLError, err:
+            print >>sys.stderr, err
+            sys.stderr.flush()
+    if tree_head_signature == None:
+        print >>sys.stderr, "Could not contact any signing nodes"
+        sys.exit(1)
+
+    sth = {"tree_size": tree_size, "timestamp": timestamp,
+           "sha256_root_hash": b64encode(root_hash),
+           "tree_head_signature": b64encode(tree_head_signature)}
+
+    check_sth_signature(ctbaseurl, sth, publickey=logpublickey)
+    timing_point(timing, "build sth")
+
+    print hexencode(root_hash), timestamp, tree_size
+    sys.stdout.flush()
+
+    write_file(sthfile, sth)
+
+    if args.timing:
+        print >>sys.stderr, timing["deltatimes"]
+        sys.stderr.flush()
+
+def main():
+    """
+    Read file 'sth' to get current tree size, assuming zero if file not
+    found.
+
+    Read tree sizes from the backup.<secondary> files, put them in a
+    list and sort it. Let new tree size equal list[backup-quorum]. Barf
+    on a new tree size smaller than the currently published tree size.
+
+    Decide on a timestamp, build an STH and write it to file 'sth'.
+    """
+    args, config, localconfig = parse_args()
+
+    while True:
+        merge_sth(args, config, localconfig)
+        if args.interval is None:
+            break
+        print >>sys.stderr, "sleeping", args.interval, "seconds"
+        time.sleep(args.interval)
+
+if __name__ == '__main__':
+    sys.exit(main())
diff --git a/tools/mergetools.py b/tools/mergetools.py
index 7644dac..86f9255 100644
--- a/tools/mergetools.py
+++ b/tools/mergetools.py
@@ -9,14 +9,43 @@ import struct
 import urllib
 import urllib2
 import json
+import yaml
+import argparse
 from certtools import get_leaf_hash, http_request, get_leaf_hash
 
 def parselogrow(row):
     return base64.b16decode(row, casefold=True)
 
-def get_logorder(filename):
-    f = open(filename, "r")
-    return [parselogrow(row.rstrip()) for row in f]
+def get_logorder(filename, items=-1):
+    logorder = []
+    n = 0
+    for row in open(filename, "r"):
+        if n == items:
+            break
+        logorder.append(parselogrow(row.rstrip()))
+        n += 1
+    return logorder
+
+def get_nfetched(currentsizefile, logorderfile):
+    try:
+        limit = json.loads(open(currentsizefile).read())
+    except (IOError, ValueError):
+        return -1
+    if limit['index'] >= 0:
+        with open(logorderfile, 'r') as f:
+            f.seek(limit['index']*65)
+            assert f.read(64).lower() == limit['hash']
+    return limit['index'] + 1
+
+def get_sth(filename):
+    try:
+        sth = json.loads(open(filename, 'r').read())
+    except (IOError, ValueError):
+        sth = {'tree_size': -1,
+               'timestamp': 0,
+               'sha256_root_hash': '',
+               'tree_head_signature': ''}
+    return sth
 
 def read_chain_open(chainsdir, filename):
     path = chainsdir + "/" + \
@@ -104,6 +133,9 @@ def verify_entry(verifycert, entry, ehash):
 def hexencode(key):
     return base64.b16encode(key).lower()
 
+def hexdecode(s):
+    return base64.b16decode(s.upper())
+
 def write_chain(key, value, chainsdir, hashed_dir=True):
     filename = hexencode(key)
     if hashed_dir:
@@ -356,3 +388,20 @@ def get_missingentriesforbackup(node, baseurl, own_key, paths):
 
 def chunks(l, n):
     return [l[i:i+n] for i in range(0, len(l), n)]
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="")
+    parser.add_argument('--config', help="System configuration",
+                        required=True)
+    parser.add_argument('--localconfig', help="Local configuration",
+                        required=True)
+    parser.add_argument('--interval', type=int, metavar="n",
+                        help="Repeate every N seconds")
+    parser.add_argument("--timing", action='store_true',
+                        help="Print timing information")
+    args = parser.parse_args()
+
+    config = yaml.load(open(args.config))
+    localconfig = yaml.load(open(args.localconfig))
+
+    return (args, config, localconfig)
diff --git a/tools/testcase1.py b/tools/testcase1.py
index c66d976..697cc99 100755
--- a/tools/testcase1.py
+++ b/tools/testcase1.py
@@ -37,7 +37,7 @@ logpublickey = get_public_key_from_file(logpublickeyfile)
 
 def testgroup(name):
     global indentation
-    print name + ":"
+    print "testgroup " + name + ":"
     indentation = "    "
 
 def print_error(message, *args):
@@ -147,7 +147,7 @@ def get_and_check_entry(timestamp, chain, leaf_index, baseurl):
                     len(submittedcertchain))
 
 def merge():
-    return subprocess.call(["../tools/merge.py", "--config", "../test/catlfish-test.cfg",
+    return subprocess.call(["../tools/merge", "--config", "../test/catlfish-test.cfg",
                             "--localconfig", "../test/catlfish-test-local-merge.cfg"])
 
 mergeresult = merge()