ceph/doc/rados/operations/read-balancer.rst

   1 .. _read_balancer:
   2
   3 =======================================
   4 Operating the Read (Primary) Balancer
   5 =======================================
   6
   7 You might be wondering: How can I improve performance in my Ceph cluster?
   8 One important data point you can check is the ``read_balance_score`` on each
   9 of your replicated pools.
  10
  11 This metric, available via ``ceph osd pool ls detail`` (see :ref:`rados_pools`
  12 for more details) indicates read performance, or how balanced the primaries are
  13 for each replicated pool. In most cases, if a ``read_balance_score`` is above 1
  14 (for instance, 1.5), this means that your pool has unbalanced primaries and that
  15 you may want to try improving your read performance with the read balancer.
  16
  17 Online Optimization
  18 ===================
  19
  20 At present, there is no online option for the read balancer. However, we plan to add
  21 the read balancer as an option to the :ref:`balancer` in the next Ceph version
  22 so it can be enabled to run automatically in the background like the upmap balancer.
  23
  24 Offline Optimization
  25 ====================
  26
  27 Primaries are updated with an offline optimizer that is built into the
  28 :ref:`osdmaptool`.
  29
  30 #. Grab the latest copy of your osdmap:
  31
  32    .. prompt:: bash $
  33
  34       ceph osd getmap -o om
  35
  36 #. Run the optimizer:
  37
  38    .. prompt:: bash $
  39
  40       osdmaptool om --read out.txt --read-pool <pool name> [--vstart]
  41
  42    It is highly recommended that you run the capacity balancer before running the
  43    balancer to ensure optimal results. See :ref:`upmap` for details on how to balance
  44    capacity in a cluster.
  45
  46 #. Apply the changes:
  47
  48    .. prompt:: bash $
  49
  50       source out.txt
  51
  52    In the above example, the proposed changes are written to the output file
  53    ``out.txt``. The commands in this procedure are normal Ceph CLI commands
  54    that can be run in order to apply the changes to the cluster.
  55
  56    If you are working in a vstart cluster, you may pass the ``--vstart`` parameter
  57    as shown above so the CLI commands are formatted with the `./bin/` prefix.
  58
  59    Note that any time the number of pgs changes (for instance, if the pg autoscaler [:ref:`pg-autoscaler`]
  60    kicks in), you should consider rechecking the scores and rerunning the balancer if needed.
  61
  62 To see some details about what the tool is doing, you can pass
  63 ``--debug-osd 10`` to ``osdmaptool``. To see even more details, pass
  64 ``--debug-osd 20`` to ``osdmaptool``.