diff --git a/_posts/2018-02-15-data-recovery.markdown b/_posts/2018-02-15-data-recovery.markdown
new file mode 100644
index 0000000..bc32e50
--- /dev/null
+++ b/_posts/2018-02-15-data-recovery.markdown
@@ -0,0 +1,41 @@
+---
+title: How can I rescued almost data from damaged HDD
+date: 2018-02-15 12:45:00 +09:00
+---
+
+This is a complete guide describes how to rescue your data from old and wrecked HDD.
+
+Suppose you are using macOS and your endangered disk is formatted with HFS+.
+
+# Beforehand
+
+## Inspect
+Use `diskutil list` to verify that which drive is what you expected for.
+This article assumes that `disk2` is the damaged one AND a partition `disk2s2` is what you expected to be rescued.
+
+## Damage Control
+To prevent extra load, unmount the damaged disk: `diskutil unmountDisk disk2`.
+
+# Rescue
+If you never been `ddrescue`, `brew install ddrescue` to install them on your machine.
+
+```bash
+cd /Volumes/AnotherDriveLargerThanDamagedDrive
+sudo ddrescue -n -v /dev/disk2s2 ./hdimage.dmg ./mapfile
+```
+
+So this command will start rescuing your data from `/dev/disk2s2` partition to `hdimage.dmg` while writing log to `mapfile`.
+You might want to rescue data as fast as possible. option `-n` is here for. This will skip scraping phase that causes aggressive disk access. 
+Option `-v` stand for verbose logging.
+
+```bash
+sudo ddrescue -r5 -v /dev/rdisk2s2 ./hdimage.dmg ./mapfile
+```
+
+When the first command completed, do it again with different parameters to aggressively scrape bad area failed to access the first time.
+Option `-r5` means ddrescue will try rescuing damaged area for 5 times.
+And `/dev/disk2s2` become `/dev/__r__disk2s2` this time. `r` stand for raw so this will access the disk more direct way.
+
+# Aftercare
+
+Mount hdimage.dmg and copy files and directories to a new drive. If the image seems broken, you can recover using `testdisk`. 
\ No newline at end of file