Log the size of each of the stores and complete DB on launch and after each

components/safe_browsing_db/v4_store.cc

Index: components/safe_browsing_db/v4_store.cc
diff --git a/components/safe_browsing_db/v4_store.cc b/components/safe_browsing_db/v4_store.cc
index 49ec3c51c711aa82d61d27aba9858de31c9c526f..7faadcd46f22029eeeb48d30241eb24ab118fc29 100644
--- a/components/safe_browsing_db/v4_store.cc
+++ b/components/safe_browsing_db/v4_store.cc
@@ -181,7 +181,7 @@ std::ostream& operator<<(std::ostream& os, const V4Store& store) {
 V4Store* V4StoreFactory::CreateV4Store(
     const scoped_refptr<base::SequencedTaskRunner>& task_runner,
     const base::FilePath& store_path) {
-  V4Store* new_store = new V4Store(task_runner, store_path);
+  V4Store* new_store = new V4Store(task_runner, store_path, 0);
   new_store->Initialize();
   return new_store;
 }
@@ -195,8 +195,11 @@ void V4Store::Initialize() {
 }
 
 V4Store::V4Store(const scoped_refptr<base::SequencedTaskRunner>& task_runner,
-                 const base::FilePath& store_path)
-    : store_path_(store_path), task_runner_(task_runner) {}
+                 const base::FilePath& store_path,
+                 const uint64_t old_file_size)
+    : file_size_(old_file_size),
+      store_path_(store_path),
+      task_runner_(task_runner) {}
 
 V4Store::~V4Store() {
   DCHECK(task_runner_->RunsTasksOnCurrentThread());
@@ -347,7 +350,8 @@ void V4Store::ApplyUpdate(
     std::unique_ptr<ListUpdateResponse> response,
     const scoped_refptr<base::SingleThreadTaskRunner>& callback_task_runner,
     UpdatedStoreReadyCallback callback) {
-  std::unique_ptr<V4Store> new_store(new V4Store(task_runner_, store_path_));
+  std::unique_ptr<V4Store> new_store(
+      new V4Store(task_runner_, store_path_, file_size_));
   ApplyUpdateResult apply_update_result;
   std::string metric;
   TimeTicks before = TimeTicks::Now();
@@ -657,6 +661,7 @@ StoreReadResult V4Store::ReadFromDisk() {
   DCHECK(task_runner_->RunsTasksOnCurrentThread());
 
   V4StoreFileFormat file_format;
+  uint64_t file_size;
   TimeTicks before = TimeTicks::Now();
   {
     // A temporary scope to make sure that |contents| get destroyed as soon as
@@ -674,6 +679,7 @@ StoreReadResult V4Store::ReadFromDisk() {
     if (!file_format.ParseFromString(contents)) {
       return PROTO_PARSING_FAILURE;
     }
+    file_size = contents.size();
   }
 
   if (file_format.magic_number() != kFileMagic) {
@@ -701,10 +707,13 @@ StoreReadResult V4Store::ReadFromDisk() {
   }
   RecordApplyUpdateTime(kReadFromDisk, TimeTicks::Now() - before, store_path_);
 
+  // Update |file_size_| now because we parsed the file correctly.
+  file_size_ = file_size;
+
   return READ_SUCCESS;
 }
 
-StoreWriteResult V4Store::WriteToDisk(const Checksum& checksum) const {
+StoreWriteResult V4Store::WriteToDisk(const Checksum& checksum) {
   V4StoreFileFormat file_format;
   ListUpdateResponse* lur = file_format.mutable_list_update_response();
   *(lur->mutable_checksum()) = checksum;
@@ -735,6 +744,9 @@ StoreWriteResult V4Store::WriteToDisk(const Checksum& checksum) const {
     return UNABLE_TO_RENAME_FAILURE;
   }
 
+  // Update |file_size_| now because we wrote the file correctly.
+  file_size_ = written;
+
   return WRITE_SUCCESS;
 }
 
@@ -832,4 +844,17 @@ bool V4Store::VerifyChecksum() {
   return true;
 }
 
+uint64_t V4Store::RecordAndReturnFileSize(const std::string& base_metric) {
+  std::string suffix = GetUmaSuffixForStore(store_path_);
+  // Histogram properties as in UMA_HISTOGRAM_COUNTS macro.
+  base::HistogramBase* histogram = base::Histogram::FactoryGet(
+      base_metric + suffix, 1, 1000000, 50,

[Nathan Parker, 2016/10/24 20:33:54]

nit: I think this would be better as '1, 10000, 50'

As is this will generate the following buckets (I think):

% python
[1.0, 1.2, 1.5, 2.0, 2.5, 3.2, 4.0, 5.1, 6.5, 8.2, 10.4, 13.2, 16.7, 21.2, 26.8,
33.9, 42.9, 54.2, 68.6, 86.8, 109.8, 138.9, 175.7, 222.2, 281.1, 355.6, 449.8,
568.9, 719.6, 910.2, 1151.3, 1456.3, 1842.0, 2329.9, 2947.0, 3727.5, 4714.8,
5963.6, 7543.1, 9540.9, 12067.9, 15264.1, 19306.9, 24420.5, 30888.4, 39069.3,
49417.1, 62505.5, 79060.4, 100000.0]

That starts to lose resolution in the relevant range of ~1MB (1151KB -> 1456
KB).  If you set the max to 100MB we'll get better resolution there:
[1.0, 1.2, 1.4, 1.7, 2.1, 2.5, 3.0, 3.7, 4.4, 5.4, 6.5, 7.9, 9.5, 11.5, 13.8,
16.7, 20.2, 24.4, 29.4, 35.5, 42.9, 51.7, 62.5, 75.4, 91.0, 109.8, 132.5, 159.9,
193.0, 232.9, 281.1, 339.3, 409.4, 494.1, 596.3, 719.6, 868.5, 1048.1, 1264.8,
1526.4, 1842.0, 2222.9, 2682.6, 3237.4, 3906.9, 4714.8, 5689.8, 6866.4, 8286.4,
10000.0]

[vakh (use Gerrit instead), 2016/10/24 22:04:51]

Unless I am looking at it incorrectly that's 10 MB.
Since the size of Soceng list is ~5MB already, I think 10 MB may be too low. How
about 50,000?
[1.0, 1.2, 1.5, 1.9, 2.4, 3.0, 3.7, 4.6, 5.8, 7.2, 9.0, 11.3, 14.1, 17.6, 22.0,
27.4, 34.2, 42.6, 53.2, 66.3, 82.7, 103.2, 128.7, 160.5, 200.2, 249.7, 311.4,
388.3, 484.3, 604.0, 753.2, 939.3, 1171.4, 1460.9, 1821.9, 2272.1, 2833.5,
3533.6, 4406.7, 5495.6, 6853.5, 8547.0, 10658.9, 13292.6, 16577.1, 20673.2,
25781.3, 32151.7, 40096.0, 50003.4]

[Scott Hess - ex-Googler, 2016/10/24 22:11:38]

Flipping the question over, do you really care about files under 100k?  50k? 
Noticing that the files are growing seems more actionable than noticing that
they are getting exceedingly small.

+      base::HistogramBase::kUmaTargetedHistogramFlag);
+  if (histogram) {
+    const int file_size_kilobytes = static_cast<int>(file_size_ / 1024);
+    histogram->Add(file_size_kilobytes);
+  }
+  return file_size_;
+}
+
 }  // namespace safe_browsing