Add L2ARC arcstats for MFU/MRU buffers and buffer content type

Currently the ARC state (MFU/MRU) of cached L2ARC buffer and their
content type is unknown. Knowing this information  may prove beneficial
in adjusting the L2ARC caching policy in the future.

This commit adds L2ARC arcstats that display the aligned size
(in bytes) of L2ARC buffers according to their content type
(data/metadata) and according to their ARC state (MRU/MFU or
prefetch). It also expands the existing evict_l2_eligible arcstat to
differentiate between MFU and MRU buffers.

L2ARC caches buffers from the MRU and MFU lists of ARC. Upon caching a
buffer, its ARC state (MRU/MFU) is stored in the L2 header
(b_arcs_state). The l2_m{f,r}u_asize arcstats reflect the aligned size
(in bytes) of L2ARC buffers according to their ARC state (based on
b_arcs_state). We also account for the case where an L2ARC and ARC
cached MRU or MRU_ghost buffer transitions to MFU. The l2_prefetch_asize
reflects the alinged size (in bytes) of L2ARC buffers that were cached
while they had the prefetch flag set in ARC. This is dynamically updated
as the prefetch flag of L2ARC buffers changes.

When buffers are evicted from ARC, if they are determined to be L2ARC
eligible then their logical size is recorded in
evict_l2_eligible_m{r,f}u arcstats according to their ARC state upon
eviction.

Persistent L2ARC:
When commiting an L2ARC buffer to a log block (L2ARC metadata) its
b_arcs_state and prefetch flag is also stored. If the buffer changes
its arcstate or prefetch flag this is reflected in the above arcstats.
However, the L2ARC metadata cannot currently be updated to reflect this
change.
Example: L2ARC caches an MRU buffer. L2ARC metadata and arcstats count
this as an MRU buffer. The buffer transitions to MFU. The arcstats are
updated to reflect this. Upon pool re-import or on/offlining the L2ARC
device the arcstats are cleared and the buffer will now be counted as an
MRU buffer, as the L2ARC metadata were not updated.

Bug fix:
- If l2arc_noprefetch is set, arc_read_done clears the L2CACHE flag of
  an ARC buffer. However, prefetches may be issued in a way that
  arc_read_done() is bypassed. Instead, move the related code in
  l2arc_write_eligible() to account for those cases too.

Signed-off-by: George Amanakis <[email protected]>

Author: gamanakis

cmd/arc_summary/arc_summary2

@@ -677,6 +677,11 @@ def get_l2arc_summary(Kstat):
     l2_writes_done = Kstat["kstat.zfs.misc.arcstats.l2_writes_done"]
     l2_writes_error = Kstat["kstat.zfs.misc.arcstats.l2_writes_error"]
     l2_writes_sent = Kstat["kstat.zfs.misc.arcstats.l2_writes_sent"]
+    l2_mfu_asize = Kstat["kstat.zfs.misc.arcstats.l2_mfu_asize"]
+    l2_mru_asize = Kstat["kstat.zfs.misc.arcstats.l2_mru_asize"]
+    l2_prefetch_asize = Kstat["kstat.zfs.misc.arcstats.l2_prefetch_asize"]
+    l2_bufc_data_asize = Kstat["kstat.zfs.misc.arcstats.l2_bufc_data_asize"]
+    l2_bufc_metadata_asize = Kstat["kstat.zfs.misc.arcstats.l2_bufc_metadata_asize"]
 
     l2_access_total = (l2_hits + l2_misses)
     output['l2_health_count'] = (l2_writes_error + l2_cksum_bad + l2_io_error)
@@ -699,7 +704,7 @@ def get_l2arc_summary(Kstat):
         output["io_errors"] = fHits(l2_io_error)
 
         output["l2_arc_size"] = {}
-        output["l2_arc_size"]["adative"] = fBytes(l2_size)
+        output["l2_arc_size"]["adaptive"] = fBytes(l2_size)
         output["l2_arc_size"]["actual"] = {
             'per': fPerc(l2_asize, l2_size),
             'num': fBytes(l2_asize)
@@ -708,6 +713,26 @@ def get_l2arc_summary(Kstat):
             'per': fPerc(l2_hdr_size, l2_size),
             'num': fBytes(l2_hdr_size),
         }
+        output["l2_arc_size"]["mfu_asize"] = {
+            'per': fPerc(l2_mfu_asize, l2_asize),
+            'num': fBytes(l2_mfu_asize),
+        }
+        output["l2_arc_size"]["mru_asize"] = {
+            'per': fPerc(l2_mru_asize, l2_asize),
+            'num': fBytes(l2_mru_asize),
+        }
+        output["l2_arc_size"]["prefetch_asize"] = {
+            'per': fPerc(l2_prefetch_asize, l2_asize),
+            'num': fBytes(l2_prefetch_asize),
+        }
+        output["l2_arc_size"]["bufc_data_asize"] = {
+            'per': fPerc(l2_bufc_data_asize, l2_asize),
+            'num': fBytes(l2_bufc_data_asize),
+        }
+        output["l2_arc_size"]["bufc_metadata_asize"] = {
+            'per': fPerc(l2_bufc_metadata_asize, l2_asize),
+            'num': fBytes(l2_bufc_metadata_asize),
+        }
 
         output["l2_arc_evicts"] = {}
         output["l2_arc_evicts"]['lock_retries'] = fHits(l2_evict_lock_retry)
@@ -772,7 +797,7 @@ def _l2arc_summary(Kstat):
         sys.stdout.write("\n")
 
         sys.stdout.write("L2 ARC Size: (Adaptive)\t\t\t\t%s\n" %
-                         arc["l2_arc_size"]["adative"])
+                         arc["l2_arc_size"]["adaptive"])
         sys.stdout.write("\tCompressed:\t\t\t%s\t%s\n" % (
             arc["l2_arc_size"]["actual"]["per"],
             arc["l2_arc_size"]["actual"]["num"],
@@ -783,6 +808,31 @@ def _l2arc_summary(Kstat):
             arc["l2_arc_size"]["head_size"]["num"],
             )
         )
+        sys.stdout.write("\tMFU Size:\t\t\t%s\t%s\n" % (
+            arc["l2_arc_size"]["mfu_asize"]["per"],
+            arc["l2_arc_size"]["mfu_asize"]["num"],
+            )
+        )
+        sys.stdout.write("\tMRU Size:\t\t\t%s\t%s\n" % (
+            arc["l2_arc_size"]["mru_asize"]["per"],
+            arc["l2_arc_size"]["mru_asize"]["num"],
+            )
+        )
+        sys.stdout.write("\tPrefetch Size:\t\t\t%s\t%s\n" % (
+            arc["l2_arc_size"]["prefetch_asize"]["per"],
+            arc["l2_arc_size"]["prefetch_asize"]["num"],
+            )
+        )
+        sys.stdout.write("\tData (buf content) Size:\t%s\t%s\n" % (
+            arc["l2_arc_size"]["bufc_data_asize"]["per"],
+            arc["l2_arc_size"]["bufc_data_asize"]["num"],
+            )
+        )
+        sys.stdout.write("\tMetadata (buf content) Size:\t%s\t%s\n" % (
+            arc["l2_arc_size"]["bufc_metadata_asize"]["per"],
+            arc["l2_arc_size"]["bufc_metadata_asize"]["num"],
+            )
+        )
         sys.stdout.write("\n")
 
         if arc["l2_arc_evicts"]['lock_retries'] != '0' or \

cmd/arc_summary/arc_summary3

@@ -736,6 +736,21 @@ def section_l2arc(kstats_dict):
     prt_i2('Header size:',
            f_perc(arc_stats['l2_hdr_size'], arc_stats['l2_size']),
            f_bytes(arc_stats['l2_hdr_size']))
+    prt_i2('MFU aligned size:',
+           f_perc(arc_stats['l2_mfu_asize'], arc_stats['l2_asize']),
+           f_bytes(arc_stats['l2_mfu_asize']))
+    prt_i2('MRU aligned size:',
+           f_perc(arc_stats['l2_mru_asize'], arc_stats['l2_asize']),
+           f_bytes(arc_stats['l2_mru_asize']))
+    prt_i2('Prefetch aligned size:',
+           f_perc(arc_stats['l2_prefetch_asize'], arc_stats['l2_asize']),
+           f_bytes(arc_stats['l2_prefetch_asize']))
+    prt_i2('Data (buffer content) aligned size:',
+           f_perc(arc_stats['l2_bufc_data_asize'], arc_stats['l2_asize']),
+           f_bytes(arc_stats['l2_bufc_data_asize']))
+    prt_i2('Metadata (buffer content) aligned size:',
+           f_perc(arc_stats['l2_bufc_metadata_asize'], arc_stats['l2_asize']),
+           f_bytes(arc_stats['l2_bufc_metadata_asize']))
 
     print()
     prt_1('L2ARC breakdown:', f_hits(l2_access_total))

cmd/zdb/zdb.c

@@ -4188,6 +4188,8 @@ dump_l2arc_log_entries(uint64_t log_entries,
 		    (u_longlong_t)L2BLK_GET_PREFETCH((&le[j])->le_prop));
 		(void) printf("|\t\t\t\taddress: %llu\n",
 		    (u_longlong_t)le[j].le_daddr);
+		(void) printf("|\t\t\t\tstate: %llu\n",
+		    (u_longlong_t)L2BLK_GET_STATE((&le[j])->le_prop));
 		(void) printf("|\n");
 	}
 	(void) printf("\n");

configure.ac

@@ -338,6 +338,7 @@ AC_CONFIG_FILES([
 	tests/zfs-tests/tests/functional/inheritance/Makefile
 	tests/zfs-tests/tests/functional/inuse/Makefile
 	tests/zfs-tests/tests/functional/io/Makefile
+	tests/zfs-tests/tests/functional/l2arc_arcstats/Makefile
 	tests/zfs-tests/tests/functional/large_files/Makefile
 	tests/zfs-tests/tests/functional/largest_pool/Makefile
 	tests/zfs-tests/tests/functional/libzfs/Makefile

include/sys/arc_impl.h

@@ -350,6 +350,8 @@ typedef struct l2arc_lb_ptr_buf {
 #define	L2BLK_SET_TYPE(field, x)	BF64_SET((field), 48, 8, x)
 #define	L2BLK_GET_PROTECTED(field)	BF64_GET((field), 56, 1)
 #define	L2BLK_SET_PROTECTED(field, x)	BF64_SET((field), 56, 1, x)
+#define	L2BLK_GET_STATE(field)		BF64_GET((field), 57, 4)
+#define	L2BLK_SET_STATE(field, x)	BF64_SET((field), 57, 4, x)
 
 #define	PTR_SWAP(x, y)		\
 	do {			\
@@ -446,6 +448,7 @@ typedef struct l2arc_buf_hdr {
 	uint64_t		b_daddr;	/* disk address, offset byte */
 	uint32_t		b_hits;
 	list_node_t		b_l2node;
+	arc_state_type_t	b_arcs_state;
 } l2arc_buf_hdr_t;
 
 typedef struct l2arc_write_callback {
@@ -546,6 +549,8 @@ typedef struct arc_stats {
 	kstat_named_t arcstat_evict_not_enough;
 	kstat_named_t arcstat_evict_l2_cached;
 	kstat_named_t arcstat_evict_l2_eligible;
+	kstat_named_t arcstat_evict_l2_eligible_mfu;
+	kstat_named_t arcstat_evict_l2_eligible_mru;
 	kstat_named_t arcstat_evict_l2_ineligible;
 	kstat_named_t arcstat_evict_l2_skip;
 	kstat_named_t arcstat_hash_elements;
@@ -744,6 +749,18 @@ typedef struct arc_stats {
 	kstat_named_t arcstat_mfu_ghost_evictable_metadata;
 	kstat_named_t arcstat_l2_hits;
 	kstat_named_t arcstat_l2_misses;
+	/*
+	 * Aligned size (in bytes) of L2ARC cached buffers by ARC state.
+	 */
+	kstat_named_t arcstat_l2_prefetch_asize;
+	kstat_named_t arcstat_l2_mru_asize;
+	kstat_named_t arcstat_l2_mfu_asize;
+	/*
+	 * Aligned size (in bytes) of L2ARC cached buffers by buffer content
+	 * type.
+	 */
+	kstat_named_t arcstat_l2_bufc_data_asize;
+	kstat_named_t arcstat_l2_bufc_metadata_asize;
 	kstat_named_t arcstat_l2_feeds;
 	kstat_named_t arcstat_l2_rw_clash;
 	kstat_named_t arcstat_l2_read_bytes;