Skip to content
Permalink
Browse files

Allow metaslab to be unloaded even when not freed from

On large systems, the memory used by loaded metaslabs can become
a concern. While range trees are a fairly efficient data structure,
on heavily fragmented pools they can still consume a significant
amount of memory. This problem is amplified when we fail to unload
metaslabs that we aren't using. Currently, we only unload a metaslab
during metaslab_sync_done; in order for that function to be called
on a given metaslab in a given txg, we have to have dirtied that
metaslab in that txg. If the dirtying was the result of an allocation,
we wouldn't be unloading it (since it wouldn't be 8 txgs since it
was selected), so in effect we only unload a metaslab during txgs
where it's being freed from.

We move the unload logic from sync_done to a new function, and
call that function on all metaslabs in a given vdev during
vdev_sync_done().

Reviewed-by: Richard Elling <Richard.Elling@RichardElling.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Paul Dagnelie <pcd@delphix.com>
Closes #8837
  • Loading branch information...
pcd1193182 authored and tonyhutter committed Jun 7, 2019
1 parent 7776650 commit ecb4cc37ba3819045faa930c64b050d463a6640b
Showing with 40 additions and 22 deletions.
  1. +1 −0 include/sys/metaslab.h
  2. +25 −22 module/zfs/metaslab.c
  3. +14 −0 module/zfs/vdev.c
@@ -50,6 +50,7 @@ int metaslab_init(metaslab_group_t *, uint64_t, uint64_t, uint64_t,
void metaslab_fini(metaslab_t *);

int metaslab_load(metaslab_t *);
void metaslab_potentially_unload(metaslab_t *, uint64_t);
void metaslab_unload(metaslab_t *);

uint64_t metaslab_allocated_space(metaslab_t *);
@@ -20,7 +20,7 @@
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2011, 2018 by Delphix. All rights reserved.
* Copyright (c) 2011, 2019 by Delphix. All rights reserved.
* Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
* Copyright (c) 2017, Intel Corporation.
*/
@@ -2949,6 +2949,30 @@ metaslab_sync(metaslab_t *msp, uint64_t txg)
dmu_tx_commit(tx);
}

void
metaslab_potentially_unload(metaslab_t *msp, uint64_t txg)
{
/*
* If the metaslab is loaded and we've not tried to load or allocate
* from it in 'metaslab_unload_delay' txgs, then unload it.
*/
if (msp->ms_loaded &&
msp->ms_disabled == 0 &&
msp->ms_selected_txg + metaslab_unload_delay < txg) {
for (int t = 1; t < TXG_CONCURRENT_STATES; t++) {
VERIFY0(range_tree_space(
msp->ms_allocating[(txg + t) & TXG_MASK]));
}
if (msp->ms_allocator != -1) {
metaslab_passivate(msp, msp->ms_weight &
~METASLAB_ACTIVE_MASK);
}

if (!metaslab_debug_unload)
metaslab_unload(msp);
}
}

/*
* Called after a transaction group has completely synced to mark
* all of the metaslab's free space as usable.
@@ -3086,27 +3110,6 @@ metaslab_sync_done(metaslab_t *msp, uint64_t txg)
*/
metaslab_recalculate_weight_and_sort(msp);

/*
* If the metaslab is loaded and we've not tried to load or allocate
* from it in 'metaslab_unload_delay' txgs, then unload it.
*/
if (msp->ms_loaded &&
msp->ms_disabled == 0 &&
msp->ms_selected_txg + metaslab_unload_delay < txg) {

for (int t = 1; t < TXG_CONCURRENT_STATES; t++) {
VERIFY0(range_tree_space(
msp->ms_allocating[(txg + t) & TXG_MASK]));
}
if (msp->ms_allocator != -1) {
metaslab_passivate(msp, msp->ms_weight &
~METASLAB_ACTIVE_MASK);
}

if (!metaslab_debug_unload)
metaslab_unload(msp);
}

ASSERT0(range_tree_space(msp->ms_allocating[txg & TXG_MASK]));
ASSERT0(range_tree_space(msp->ms_freeing));
ASSERT0(range_tree_space(msp->ms_freed));
@@ -3234,6 +3234,20 @@ vdev_sync_done(vdev_t *vd, uint64_t txg)
!= NULL)
metaslab_sync_done(msp, txg);

/*
* Because this function is only called on dirty vdevs, it's possible
* we won't consider all metaslabs for unloading on every
* txg. However, unless the system is largely idle it is likely that
* we will dirty all vdevs within a few txgs.
*/
for (int i = 0; i < vd->vdev_ms_count; i++) {
msp = vd->vdev_ms[i];
mutex_enter(&msp->ms_lock);
if (msp->ms_sm != NULL)
metaslab_potentially_unload(msp, txg);
mutex_exit(&msp->ms_lock);
}

if (reassess)
metaslab_sync_reassess(vd->vdev_mg);
}

0 comments on commit ecb4cc3

Please sign in to comment.
You can’t perform that action at this time.