Merge remote-tracking branch 'refs/remotes/map/storagegc2'

1 files changed, 22 insertions, 3 deletions

diff --git a/tools/storagegc.py b/tools/storagegc.py
index f18074a..a4f15b7 100755
--- a/tools/storagegc.py
+++ b/tools/storagegc.py
@@ -22,7 +22,7 @@ localconfig = yaml.load(open(args.localconfig))
 
 paths = localconfig["paths"]
 db_path = paths["db"]
-create_ssl_context(cafile=paths["https_cacertfile"])
+create_ssl_context(cafile=paths.get("public_cacertfile", None))
 
 baseurl = config["baseurl"]
 
@@ -50,15 +50,34 @@ def verifyleafhash(leaf_hash):
 
 starttime = datetime.datetime.now()
 
-lastverified = (-1, None)
 
 try:
-    for i, line in enumerate(open(db_path + "newentries")):
+    lastverifiedstring = open(db_path + "lastverifiednewentry").read()
+    lastverified = json.loads(lastverifiedstring)
+except IOError:
+    lastverified = {"index": -1, "hash": None}
+print "starting at", lastverified
+
+newentriesfile = open(db_path + "newentries")
+if lastverified["index"] >= 0:
+    newentriesfile.seek(lastverified["index"]*65)
+    assert(newentriesfile.read(64).lower() == lastverified["hash"])
+newentriesfile.seek((lastverified["index"]+1)*65)
+
+try:
+    i = lastverified["index"] + 1
+    sincewritten = 0
+    for line in newentriesfile:
         leaf_hash = base64.b16decode(line.strip(), casefold=True)
         result = verifyleafhash(leaf_hash)
         if not result:
             break
         lastverified = {"index": i, "hash": base64.b16encode(leaf_hash).lower()}
+        i += 1
+        sincewritten += 1
+        if sincewritten > 1000:
+            write_file(db_path + "lastverifiednewentry", lastverified)
+            sincewritten = 0
     if lastverified["index"] >= 0:
         write_file(db_path + "lastverifiednewentry", lastverified)
     print "lastverified", lastverified