From 5d3ad39f06c28f17e2fa051c564fb2e0578ec02d Mon Sep 17 00:00:00 2001 From: Zac Dover Date: Mon, 29 Jul 2024 21:17:44 +1000 Subject: [PATCH] doc/cephfs: edit "Dynamic Subtree Partitioning" Edit the section "Dynamic Subtree Partitioning" in doc/cephfs/multimds.rst. Co-authored-by: Anthony D'Atri Signed-off-by: Zac Dover (cherry picked from commit d14119e37e01bd06eed668d9e3606a66ef205b30) --- doc/cephfs/multimds.rst | 49 ++++++++++++++++++++++------------------- 1 file changed, 26 insertions(+), 23 deletions(-) diff --git a/doc/cephfs/multimds.rst b/doc/cephfs/multimds.rst index 9855fa5e725bf..3d7a4bc8a0613 100644 --- a/doc/cephfs/multimds.rst +++ b/doc/cephfs/multimds.rst @@ -248,42 +248,45 @@ Dynamic Subtree Partitioning CephFS has long had a dynamic metadata balancer (sometimes called the "default balancer") which can split or merge subtrees while placing them on "colder" MDS -ranks. Moving the metadata around can improve overall file system throughput +ranks. Moving the metadata in this way improves overall file system throughput and cache size. -However, the balancer has suffered from problem with efficiency and performance -so it is by default turned off. This is to avoid an administrator "turning on -multimds" by increasing the ``max_mds`` setting and then finding the balancer -has made a mess of the cluster performance (reverting is straightforward but -can take time). +However, the balancer is sometimes inefficient or slow, so by default it is +turned off. This is to avoid an administrator "turning on multimds" by +increasing the ``max_mds`` setting only to find that the balancer has made a +mess of the cluster performance (reverting from this messy state of affairs is +straightforward but can take time). -The setting to turn on the balancer is: +To turn on the balancer, run a command of the following form: .. prompt:: bash # ceph fs set balance_automate true -Turning on the balancer should only be done with appropriate configuration, -such as with the ``bal_rank_mask`` setting (described below). Careful -monitoring of the file system performance and MDS is advised. +Turn on the balancer only with an appropriate configuration, such as a +configuration that includes the ``bal_rank_mask`` setting (described +:ref:`below `). + +Careful monitoring of the file system performance and MDS is advised. Dynamic subtree partitioning with Balancer on specific ranks ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -The CephFS file system provides the ``bal_rank_mask`` option to enable the balancer -to dynamically rebalance subtrees within particular active MDS ranks. This -allows administrators to employ both the dynamic subtree partitioning and -static pining schemes in different active MDS ranks so that metadata loads -are optimized based on user demand. For instance, in realistic cloud -storage environments, where a lot of subvolumes are allotted to multiple -computing nodes (e.g., VMs and containers), some subvolumes that require -high performance are managed by static partitioning, whereas most subvolumes -that experience a moderate workload are managed by the balancer. As the balancer -evenly spreads the metadata workload to all active MDS ranks, performance of -static pinned subvolumes inevitably may be affected or degraded. If this option -is enabled, subtrees managed by the balancer are not affected by -static pinned subtrees. +.. _bal-rank-mask: + +The CephFS file system provides the ``bal_rank_mask`` option to enable the +balancer to dynamically rebalance subtrees within particular active MDS ranks. +This allows administrators to employ both the dynamic subtree partitioning and +static pining schemes in different active MDS ranks so that metadata loads are +optimized based on user demand. For instance, in realistic cloud storage +environments, where a lot of subvolumes are allotted to multiple computing +nodes (e.g., VMs and containers), some subvolumes that require high performance +are managed by static partitioning, whereas most subvolumes that experience a +moderate workload are managed by the balancer. As the balancer evenly spreads +the metadata workload to all active MDS ranks, performance of static pinned +subvolumes inevitably may be affected or degraded. If this option is enabled, +subtrees managed by the balancer are not affected by static pinned subtrees. This option can be configured with the ``ceph fs set`` command. For example: -- 2.39.5