Archive-Team
/
zfs
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases 103 Wiki Activity

Reduce number of arc_prune threads 

On FreeBSD vnode reclamation is single-threaded, protected by single
global lock.  Linux seems to be able to use a thread per mount point,
but at this time it creates more harm than good.

Reduce number of threads to 1, adding tunable in case somebody wants
to try more.

Reviewed-by: Ryan Moeller <ryan@ixsystems.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Chris Dunlop <chris@onthe.net.au>
Reviewed-by: Ahelenia Ziemiańska <nabijaczleweli@nabijaczleweli.xyz>
Signed-off-by: Alexander Motin <mav@FreeBSD.org>
Closes #12896
Issue #9966
This commit is contained in:
Alexander Motin 2021-12-22 20:07:13 -05:00 committed by Tony Hutter
parent 913ae45218
commit 786abf5321
2 changed files with 17 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
man/man4/zfs.4
View File

@ -671,6 +671,13 @@ Minimum time "prescient prefetched" blocks are locked in the ARC.
These blocks are meant to be prefetched fairly aggressively ahead of These blocks are meant to be prefetched fairly aggressively ahead of
the code that may use them. the code that may use them.
. .
.It Sy zfs_arc_prune_task_threads Ns = Ns Sy 1 Pq int
Number of arc_prune threads.
.Fx
does not need more than one.
Linux may theoretically use one per mount point up to number of CPUs,
but that was not proven to be useful.
.
.It Sy zfs_max_missing_tvds Ns = Ns Sy 0 Pq int .It Sy zfs_max_missing_tvds Ns = Ns Sy 0 Pq int
Number of missing top-level vdevs which will be allowed during Number of missing top-level vdevs which will be allowed during
pool import (only in read-only mode). pool import (only in read-only mode).

13
module/zfs/arc.c
View File

@ -463,6 +463,11 @@ int zfs_arc_meta_strategy = ARC_STRATEGY_META_BALANCED;
int zfs_arc_meta_adjust_restarts = 4096; int zfs_arc_meta_adjust_restarts = 4096;
int zfs_arc_lotsfree_percent = 10; int zfs_arc_lotsfree_percent = 10;
/*
* Number of arc_prune threads
*/
static int zfs_arc_prune_task_threads = 1;
/* The 6 states: */ /* The 6 states: */
arc_state_t ARC_anon; arc_state_t ARC_anon;
arc_state_t ARC_mru; arc_state_t ARC_mru;
@ -7972,9 +7977,8 @@ arc_init(void)
offsetof(arc_prune_t, p_node)); offsetof(arc_prune_t, p_node));
mutex_init(&arc_prune_mtx, NULL, MUTEX_DEFAULT, NULL); mutex_init(&arc_prune_mtx, NULL, MUTEX_DEFAULT, NULL);
arc_prune_taskq = taskq_create("arc_prune", 100, defclsyspri, arc_prune_taskq = taskq_create("arc_prune", zfs_arc_prune_task_threads,
boot_ncpus, INT_MAX, TASKQ_PREPOPULATE | TASKQ_DYNAMIC | defclsyspri, 100, INT_MAX, TASKQ_PREPOPULATE | TASKQ_DYNAMIC);
TASKQ_THREADS_CPU_PCT);
arc_ksp = kstat_create("zfs", 0, "arcstats", "misc", KSTAT_TYPE_NAMED, arc_ksp = kstat_create("zfs", 0, "arcstats", "misc", KSTAT_TYPE_NAMED,
sizeof (arc_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL); sizeof (arc_stats) / sizeof (kstat_named_t), KSTAT_FLAG_VIRTUAL);
@ -11106,4 +11110,7 @@ ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, eviction_pct, INT, ZMOD_RW,
ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, evict_batch_limit, INT, ZMOD_RW, ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, evict_batch_limit, INT, ZMOD_RW,
"The number of headers to evict per sublist before moving to the next"); "The number of headers to evict per sublist before moving to the next");
ZFS_MODULE_PARAM(zfs_arc, zfs_arc_, prune_task_threads, INT, ZMOD_RW,
"Number of arc_prune threads");
/* END CSTYLED */ /* END CSTYLED */