Record the hardlinks in the commit message, so we can restore them

git doesn't care about hardlinks; when it does the checkout it creates
different files.

Here we record the hardlinks in the install_dir as a json blob in the
commit message, then use that to re-hardlink after cache restore.

Signed-off-by: Richard Clamp <[email protected]>

lib/omnibus/git_cache.rb

@@ -32,7 +32,7 @@ class GitCache
     # will not have the generated content, so these snapshots would be
     # incompatible with the current omnibus codebase. Incrementing the serial
     # number ensures these old shapshots will not be used in subsequent builds.
-    SERIAL_NUMBER = 3
+    SERIAL_NUMBER = 4
 
     REQUIRED_GIT_FILES = %w{
 HEAD
@@ -126,11 +126,12 @@ def incremental
 
       create_cache_path
       remove_git_dirs
+      hardlinks = find_hardlinks
 
       git_cmd("add -A -f")
 
       begin
-        git_cmd(%Q{commit -q -m "Backup of #{tag}"})
+        git_cmd("commit -q -F -", input: "Backup of #{tag}\n\n#{FFI_Yajl::Encoder.encode(hardlinks, pretty: true)}")
       rescue CommandFailed => e
         raise unless e.message.include?("nothing to commit")
       end
@@ -159,6 +160,7 @@ def restore
 
     def restore_from_cache
       git_cmd("checkout -f restore_here")
+      restore_hardlinks
     ensure
       git_cmd("tag -d restore_here")
     end
@@ -184,6 +186,43 @@ def remove_git_dirs
       true
     end
 
+    # Discover any hardlinked files in the install_dir
+    #
+    # @return [Hash{String => Array<String>}]
+    def find_hardlinks
+      hardlink_sources = {}
+      hardlinks = {}
+      Omnibus::FileSyncer.all_files_under(install_dir).each do |path|
+        stat = File.stat(path)
+        if stat.ftype.to_sym == :file && stat.nlink > 1
+          key = [stat.dev, stat.ino]
+          if source = hardlink_sources[key]
+            hardlinks[source] ||= []
+            hardlinks[source] << path
+          else
+            hardlink_sources[key] = path
+          end
+        end
+      end
+      hardlinks
+    end
+
+    # Restores any hardlinking from the commit message body.  Body is assumed to
+    # be the JSON encoded return value from #find_hardlinks.
+    #
+    # @return true
+    def restore_hardlinks
+      body = git_cmd("log --format=%b -n 1").stdout
+      hardlinks = FFI_Yajl::Parser.parse(body)
+
+      hardlinks.each do |source, dest|
+        dest.each do |path|
+          FileUtils.ln(source, path, force: true)
+        end
+      end
+      true
+    end
+
     private
 
     #

spec/unit/git_cache_spec.rb

@@ -105,10 +105,83 @@ module Omnibus
       end
     end
 
+    describe "#find_hardlinks" do
+      let(:regular_file_stat) do
+        stat = double(File::Stat)
+        allow(stat).to receive(:ftype).and_return(:file)
+        allow(stat).to receive(:nlink).and_return(1)
+        stat
+      end
+
+      let(:hardlinked_file_stat) do
+        stat = double(File::Stat)
+        allow(stat).to receive(:ftype).and_return(:file)
+        allow(stat).to receive(:nlink).and_return(2)
+        allow(stat).to receive(:dev).and_return(5)
+        allow(stat).to receive(:ino).and_return(25)
+        stat
+      end
+
+      before do
+        allow(File).to receive(:stat).and_return(regular_file_stat)
+        allow(File).to receive(:stat).with("foo").and_return(hardlinked_file_stat)
+        allow(File).to receive(:stat).with("bar").and_return(hardlinked_file_stat)
+
+        allow(Omnibus::FileSyncer).to receive(:all_files_under).and_return(
+          %w{ foo bar baz quux }
+        )
+      end
+
+      it "returns some hardlinks" do
+        expect(ipc.find_hardlinks).to eq({ "foo" => ["bar"] })
+      end
+    end
+
+    describe "#restore_hardlinks" do
+      let(:hardlinks) do
+        {
+          "/opt/demo/bin/file1" => [
+            "/opt/demo/bin/file2",
+            "/opt/demo/bin/file3",
+          ],
+        }
+      end
+
+      let(:git_log_output) { FFI_Yajl::Encoder.encode(hardlinks) }
+
+      let(:log_cmd) do
+        cmd_double = double(Mixlib::ShellOut)
+        allow(cmd_double).to receive(:stdout).and_return(git_log_output)
+        cmd_double
+      end
+
+      before(:each) do
+        allow(ipc).to receive(:git_cmd)
+          .with("log --format=%b -n 1").and_return(log_cmd)
+        allow(FileUtils).to receive(:ln)
+      end
+
+      it "checks the commit message" do
+        expect(ipc).to receive(:git_cmd)
+          .with("log --format=%b -n 1")
+
+        ipc.restore_hardlinks
+      end
+
+      it "recreates hardlinks" do
+        expect(FileUtils).to receive(:ln)
+          .with("/opt/demo/bin/file1", "/opt/demo/bin/file2", force: true)
+        expect(FileUtils).to receive(:ln)
+          .with("/opt/demo/bin/file1", "/opt/demo/bin/file3", force: true)
+        ipc.restore_hardlinks
+      end
+    end
+
     describe "#incremental" do
       before(:each) do
         allow(ipc).to receive(:git_cmd)
         allow(ipc).to receive(:create_cache_path)
+        allow(ipc).to receive(:find_hardlinks).and_return({})
       end
 
       it "creates the cache path" do
@@ -125,7 +198,7 @@ module Omnibus
 
       it "commits the backup for the software" do
         expect(ipc).to receive(:git_cmd)
-          .with(%Q{commit -q -m "Backup of #{ipc.tag}"})
+          .with("commit -q -F -", input: "Backup of #{ipc.tag}\n\n{\n\n}\n")
         ipc.incremental
       end
 
@@ -176,6 +249,7 @@ module Omnibus
         allow(ipc).to receive(:git_cmd)
           .with(%Q{tag -f restore_here "#{ipc.tag}"})
         allow(ipc).to receive(:create_cache_path)
+        allow(ipc).to receive(:restore_hardlinks)
       end
 
       it "creates the cache path" do