ceph/doc/rados/configuration/journal-ref.rst

   1 ==========================
   2  Journal Config Reference
   3 ==========================
   4
   5 .. index:: journal; journal configuration
   6
   7 Ceph OSDs use a journal for two reasons: speed and consistency.
   8
   9 - **Speed:** The journal enables the Ceph OSD Daemon to commit small writes
  10   quickly. Ceph writes small, random i/o to the journal sequentially, which
  11   tends to speed up bursty workloads by allowing the backing filesystem more
  12   time to coalesce writes. The Ceph OSD Daemon's journal, however, can lead
  13   to spiky performance with short spurts of high-speed writes followed by
  14   periods without any write progress as the filesystem catches up to the
  15   journal.
  16
  17 - **Consistency:** Ceph OSD Daemons require a filesystem interface that
  18   guarantees atomic compound operations. Ceph OSD Daemons write a description
  19   of the operation to the journal and apply the operation to the filesystem.
  20   This enables atomic updates to an object (for example, placement group
  21   metadata). Every few seconds--between ``filestore max sync interval`` and
  22   ``filestore min sync interval``--the Ceph OSD Daemon stops writes and
  23   synchronizes the journal with the filesystem, allowing Ceph OSD Daemons to
  24   trim operations from the journal and reuse the space. On failure, Ceph
  25   OSD Daemons replay the journal starting after the last synchronization
  26   operation.
  27
  28 Ceph OSD Daemons support the following journal settings:
  29
  30
  31 ``journal dio``
  32
  33 :Description: Enables direct i/o to the journal. Requires ``journal block
  34               align`` set to ``true``.
  35
  36 :Type: Boolean
  37 :Required: Yes when using ``aio``.
  38 :Default: ``true``
  39
  40
  41
  42 ``journal aio``
  43
  44 .. versionchanged:: 0.61 Cuttlefish
  45
  46 :Description: Enables using ``libaio`` for asynchronous writes to the journal.
  47               Requires ``journal dio`` set to ``true``.
  48
  49 :Type: Boolean
  50 :Required: No.
  51 :Default: Version 0.61 and later, ``true``. Version 0.60 and earlier, ``false``.
  52
  53
  54 ``journal block align``
  55
  56 :Description: Block aligns write operations. Required for ``dio`` and ``aio``.
  57 :Type: Boolean
  58 :Required: Yes when using ``dio`` and ``aio``.
  59 :Default: ``true``
  60
  61
  62 ``journal max write bytes``
  63
  64 :Description: The maximum number of bytes the journal will write at
  65               any one time.
  66
  67 :Type: Integer
  68 :Required: No
  69 :Default: ``10 << 20``
  70
  71
  72 ``journal max write entries``
  73
  74 :Description: The maximum number of entries the journal will write at
  75               any one time.
  76
  77 :Type: Integer
  78 :Required: No
  79 :Default: ``100``
  80
  81
  82 ``journal queue max ops``
  83
  84 :Description: The maximum number of operations allowed in the queue at
  85               any one time.
  86
  87 :Type: Integer
  88 :Required: No
  89 :Default: ``500``
  90
  91
  92 ``journal queue max bytes``
  93
  94 :Description: The maximum number of bytes allowed in the queue at
  95               any one time.
  96
  97 :Type: Integer
  98 :Required: No
  99 :Default: ``10 << 20``
 100
 101
 102 ``journal align min size``
 103
 104 :Description: Align data payloads greater than the specified minimum.
 105 :Type: Integer
 106 :Required: No
 107 :Default: ``64 << 10``
 108
 109
 110 ``journal zero on create``
 111
 112 :Description: Causes the file store to overwrite the entire journal with
 113               ``0``'s during ``mkfs``.
 114 :Type: Boolean
 115 :Required: No
 116 :Default: ``false``