)]}'
{
  "commit": "a5eaeb8670b0b56f884fbda8ddb92ea0bf78fb5d",
  "tree": "9cf2ea8bcfb5ee1170f4055979a69a0cd4266ab9",
  "parents": [
    "f042e6f95bb7dc771bf79f309dbdf0b34da933da"
  ],
  "author": {
    "name": "Lorenz Brun",
    "email": "lorenz@nexantic.com",
    "time": "Thu Jul 16 15:06:50 2020 +0200"
  },
  "committer": {
    "name": "Lorenz Brun",
    "email": "lorenz@nexantic.com",
    "time": "Thu Jul 16 15:06:50 2020 +0200"
  },
  "message": "Increase zeroing performance for data partition\n\nIt\u0027s possible to increase queue depth for O_DIRECT in simple cases like ours by just\nsubmitting bigger buffers. As long as they are an exact multiple of the block size this is fine\nand the kernel doesn\u0027t complain. This also enables O_SYNC to prevent any buffering on the guest.\nThis should help to push out data quicker and prevent buffer bloat. The host has its own cache\nanyways.\n\nTest Plan:\nNo change in functionality, I observe more predictable performance (previously\nI sometimes had stalls where the initialization would take \u003e 60s).\n\nX-Origin-Diff: phab/D599\nGitOrigin-RevId: 19554fd9e6d709bde738d01a0d2de190c441640e\n",
  "tree_diff": [
    {
      "type": "modify",
      "old_id": "fc556e1e38672dadf34944b9b7b0f34c207953e1",
      "old_mode": 33188,
      "old_path": "core/internal/storage/blockdev.go",
      "new_id": "660abe6c63ac700b145f8a84d43a1b90b9f8a157",
      "new_mode": 33188,
      "new_path": "core/internal/storage/blockdev.go"
    }
  ]
}