Allow "arc_p" to drop to zero or grow to "arc_c"

Setting a limit on the minimum value of "arc_p" has been shown to have
detrimental effects on the arc hit rate for certain "metadata" intensive
workloads. Specifically, this has been exhibited with a workload that
constantly dirties new "metadata" but also frequently touches a "small"
amount of mfu data (e.g. mkdir's).

What is seen is that the new anon data throttles the mfu list to a
negligible size (because arc_p > anon + mru in arc_get_data_buf), even
though the mfu ghost list receives a constant stream of hits. To remedy
this, arc_p is now allowed to drop to zero if the algorithm deems it
necessary.

Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Issue #2110

diff --git a/man/man5/zfs-module-parameters.5 b/man/man5/zfs-module-parameters.5
index 7f010c6cfc5bb246e64ac288b09651d8dd2f06ee..2acf68383f7273498fdcdb1f088be81319a6578a 100644 (file)
--- a/man/man5/zfs-module-parameters.5
+++ b/man/man5/zfs-module-parameters.5
@@ -293,17 +293,6 @@ Min life of prefetch block
 Default value: \fB100\fR.
 .RE
 
-.sp
-.ne 2
-.na
-\fBzfs_arc_p_min_shift\fR (int)
-.ad
-.RS 12n
-arc_c shift to calc min/max arc_p
-.sp
-Default value: \fB4\fR.
-.RE
-
 .sp
 .ne 2
 .na

diff --git a/module/zfs/arc.c b/module/zfs/arc.c
index eac6ea4488fc5569ba6dee5fa61268ce73da9095..f66eaa407096e1a84cf60059b2df80305d53ab4c 100644 (file)
--- a/module/zfs/arc.c
+++ b/module/zfs/arc.c
@@ -172,9 +172,6 @@ int arc_evict_iterations = 100;
 /* number of seconds before growing cache again */
 int zfs_arc_grow_retry = 5;
 
-/* shift of arc_c for calculating both min and max arc_p */
-int zfs_arc_p_min_shift = 4;
-
 /* disable anon data aggressively growing arc_p */
 int zfs_arc_p_aggressive_disable = 1;
 
@@ -2335,7 +2332,6 @@ void
 arc_shrink(uint64_t bytes)
 {
        if (arc_c > arc_c_min) {
-               uint64_t arc_p_min;
                uint64_t to_free;
 
                to_free = bytes ? bytes : arc_c >> zfs_arc_shrink_shift;
@@ -2345,13 +2341,12 @@ arc_shrink(uint64_t bytes)
                else
                        arc_c = arc_c_min;
 
-               arc_p_min = (arc_c >> zfs_arc_p_min_shift);
                to_free = bytes ? bytes : arc_p >> zfs_arc_shrink_shift;
 
-               if (arc_p > arc_p_min + to_free)
+               if (arc_p > to_free)
                        atomic_add_64(&arc_p, -to_free);
                else
-                       arc_p = arc_p_min;
+                       arc_p = 0;
 
                if (arc_c > arc_size)
                        arc_c = MAX(arc_size, arc_c_min);
@@ -2622,7 +2617,6 @@ static void
 arc_adapt(int bytes, arc_state_t *state)
 {
        int mult;
-       uint64_t arc_p_min = (arc_c >> zfs_arc_p_min_shift);
 
        if (state == arc_l2c_only)
                return;
@@ -2641,7 +2635,7 @@ arc_adapt(int bytes, arc_state_t *state)
                    1 : (arc_mfu_ghost->arcs_size/arc_mru_ghost->arcs_size));
                mult = MIN(mult, 10); /* avoid wild arc_p adjustment */
 
-               arc_p = MIN(arc_c - arc_p_min, arc_p + bytes * mult);
+               arc_p = MIN(arc_c, arc_p + bytes * mult);
        } else if (state == arc_mfu_ghost) {
                uint64_t delta;
 
@@ -2650,7 +2644,7 @@ arc_adapt(int bytes, arc_state_t *state)
                mult = MIN(mult, 10);
 
                delta = MIN(bytes * mult, arc_p);
-               arc_p = MAX(arc_p_min, arc_p - delta);
+               arc_p = MAX(0, arc_p - delta);
        }
        ASSERT((int64_t)arc_p >= 0);
 
@@ -5563,9 +5557,6 @@ MODULE_PARM_DESC(zfs_arc_p_aggressive_disable, "disable aggressive arc_p grow");
 module_param(zfs_arc_shrink_shift, int, 0644);
 MODULE_PARM_DESC(zfs_arc_shrink_shift, "log2(fraction of arc to reclaim)");
 
-module_param(zfs_arc_p_min_shift, int, 0644);
-MODULE_PARM_DESC(zfs_arc_p_min_shift, "arc_c shift to calc min/max arc_p");
-
 module_param(zfs_disable_dup_eviction, int, 0644);
 MODULE_PARM_DESC(zfs_disable_dup_eviction, "disable duplicate buffer eviction");