Add new kstat for monitoring time in dmu_tx_assign

This change adds a new kstat to gain some visibility into the amount of
time spent in each call to dmu_tx_assign. A histogram is exported via
a new dmu_tx_assign_histogram-$POOLNAME file. The information contained
in this histogram is the frequency dmu_tx_assign took to complete given
an interval range. For example, given the below histogram file:

    $ cat /proc/spl/kstat/zfs/dmu_tx_assign_histogram-tank
    12 1 0x01 32 1536 19792068076691 20516481514522
    name                            type data
    1 us                            4    859
    2 us                            4    252
    4 us                            4    171
    8 us                            4    2
    16 us                           4    0
    32 us                           4    2
    64 us                           4    0
    128 us                          4    0
    256 us                          4    0
    512 us                          4    0
    1024 us                         4    0
    2048 us                         4    0
    4096 us                         4    0
    8192 us                         4    0
    16384 us                        4    0
    32768 us                        4    1
    65536 us                        4    1
    131072 us                       4    1
    262144 us                       4    4
    524288 us                       4    0
    1048576 us                      4    0
    2097152 us                      4    0
    4194304 us                      4    0
    8388608 us                      4    0
    16777216 us                     4    0
    33554432 us                     4    0
    67108864 us                     4    0
    134217728 us                    4    0
    268435456 us                    4    0
    536870912 us                    4    0
    1073741824 us                   4    0
    2147483648 us                   4    0

one can see most calls to dmu_tx_assign completed in 32us or less, but a
few outliers did not. Specifically, 4 of the calls took between 262144us
and 131072us. This information is difficult, if not impossible, to gather
without this change.

Signed-off-by: Prakash Surya <surya1@llnl.gov>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #1584

include/sys/dsl_pool.h

@@ -89,6 +89,7 @@ typedef struct dsl_pool {
 	uint64_t dp_root_dir_obj;
 	struct taskq *dp_iput_taskq;
 	kstat_t *dp_txg_kstat;
+	kstat_t *dp_tx_assign_kstat;
 
 	/* No lock needed - sync context only */
 	blkptr_t dp_meta_rootbp;
@@ -111,6 +112,8 @@ typedef struct dsl_pool {
 	uint64_t dp_mos_uncompressed_delta;
 	uint64_t dp_txg_history_size;
 	list_t dp_txg_history;
+	uint64_t dp_tx_assign_size;
+	kstat_named_t *dp_tx_assign_buckets;
 
 
 	/* Has its own locking */
@@ -162,6 +165,8 @@ extern int dsl_pool_user_release(dsl_pool_t *dp, uint64_t dsobj,
 extern void dsl_pool_clean_tmp_userrefs(dsl_pool_t *dp);
 int dsl_pool_open_special_dir(dsl_pool_t *dp, const char *name, dsl_dir_t **);
 
+void dsl_pool_tx_assign_add_usecs(dsl_pool_t *dp, uint64_t usecs);
+
 txg_history_t *dsl_pool_txg_history_add(dsl_pool_t *dp, uint64_t txg);
 txg_history_t *dsl_pool_txg_history_get(dsl_pool_t *dp, uint64_t txg);
 void dsl_pool_txg_history_put(txg_history_t *th);

module/zfs/dmu_tx.c

@@ -1089,12 +1089,15 @@ dmu_tx_unassign(dmu_tx_t *tx)
 int
 dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how)
 {
+	hrtime_t before, after;
 	int err;
 
 	ASSERT(tx->tx_txg == 0);
 	ASSERT(txg_how != 0);
 	ASSERT(!dsl_pool_sync_context(tx->tx_pool));
 
+	before = gethrtime();
+
 	while ((err = dmu_tx_try_assign(tx, txg_how)) != 0) {
 		dmu_tx_unassign(tx);
 
@@ -1106,6 +1109,11 @@ dmu_tx_assign(dmu_tx_t *tx, uint64_t txg_how)
 
 	txg_rele_to_quiesce(&tx->tx_txgh);
 
+	after = gethrtime();
+
+	dsl_pool_tx_assign_add_usecs(tx->tx_pool,
+	    (after - before) / NSEC_PER_USEC);
+
 	return (0);
 }
 

module/zfs/dsl_pool.c

@@ -58,6 +58,63 @@ kmutex_t zfs_write_limit_lock;
 
 static pgcnt_t old_physmem = 0;
 
+static void
+dsl_pool_tx_assign_init(dsl_pool_t *dp, unsigned int ndata)
+{
+	kstat_named_t *ks;
+	char name[KSTAT_STRLEN];
+	int i, data_size = ndata * sizeof(kstat_named_t);
+
+	(void) snprintf(name, KSTAT_STRLEN, "dmu_tx_assign-%s",
+			spa_name(dp->dp_spa));
+
+	dp->dp_tx_assign_size = ndata;
+
+	if (data_size)
+		dp->dp_tx_assign_buckets = kmem_alloc(data_size, KM_SLEEP);
+	else
+		dp->dp_tx_assign_buckets = NULL;
+
+	for (i = 0; i < dp->dp_tx_assign_size; i++) {
+		ks = &dp->dp_tx_assign_buckets[i];
+		ks->data_type = KSTAT_DATA_UINT64;
+		ks->value.ui64 = 0;
+		(void) snprintf(ks->name, KSTAT_STRLEN, "%u us", 1 << i);
+	}
+
+	dp->dp_tx_assign_kstat = kstat_create("zfs", 0, name, "misc",
+	    KSTAT_TYPE_NAMED, 0, KSTAT_FLAG_VIRTUAL);
+
+	if (dp->dp_tx_assign_kstat) {
+		dp->dp_tx_assign_kstat->ks_data = dp->dp_tx_assign_buckets;
+		dp->dp_tx_assign_kstat->ks_ndata = dp->dp_tx_assign_size;
+		dp->dp_tx_assign_kstat->ks_data_size = data_size;
+		kstat_install(dp->dp_tx_assign_kstat);
+	}
+}
+
+static void
+dsl_pool_tx_assign_destroy(dsl_pool_t *dp)
+{
+	if (dp->dp_tx_assign_buckets)
+		kmem_free(dp->dp_tx_assign_buckets,
+			  dp->dp_tx_assign_size * sizeof(kstat_named_t));
+
+	if (dp->dp_tx_assign_kstat)
+		kstat_delete(dp->dp_tx_assign_kstat);
+}
+
+void
+dsl_pool_tx_assign_add_usecs(dsl_pool_t *dp, uint64_t usecs)
+{
+	uint64_t idx = 0;
+
+	while (((1 << idx) < usecs) && (idx < dp->dp_tx_assign_size - 1))
+		idx++;
+
+	atomic_inc_64(&dp->dp_tx_assign_buckets[idx].value.ui64);
+}
+
 static int
 dsl_pool_txg_history_update(kstat_t *ksp, int rw)
 {
@@ -238,6 +295,7 @@ dsl_pool_open_impl(spa_t *spa, uint64_t txg)
 	    1, 4, 0);
 
 	dsl_pool_txg_history_init(dp, txg);
+	dsl_pool_tx_assign_init(dp, 32);
 
 	return (dp);
 }
@@ -379,6 +437,7 @@ dsl_pool_close(dsl_pool_t *dp)
 	arc_flush(dp->dp_spa);
 	txg_fini(dp);
 	dsl_scan_fini(dp);
+	dsl_pool_tx_assign_destroy(dp);
 	dsl_pool_txg_history_destroy(dp);
 	rw_destroy(&dp->dp_config_rwlock);
 	mutex_destroy(&dp->dp_lock);