Introduce read/write kstats per dataset

The following patch introduces a few statistics on reads and writes
grouped by dataset. These statistics are implemented as kstats
(backed by aggregate sums for performance) and can be retrieved by
using the dataset objset ID number. The motivation for this change is
to provide some preliminary analytics on dataset usage/performance.

Reviewed-by: Richard Elling <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Reviewed by: Matthew Ahrens <[email protected]>
Signed-off-by: Serapheim Dimitropoulos <[email protected]>
Closes #7705 

1 files changed, 185 insertions, 0 deletions

diff --git a/module/zfs/dataset_kstats.c b/module/zfs/dataset_kstats.c
new file mode 100644
index 000000000..ac0ad84ed
--- /dev/null
+++ b/module/zfs/dataset_kstats.c
@@ -0,0 +1,185 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License (the "License").
+ * You may not use this file except in compliance with the License.
+ *
+ * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+ * or http://www.opensolaris.org/os/licensing.
+ * See the License for the specific language governing permissions
+ * and limitations under the License.
+ *
+ * When distributing Covered Code, include this CDDL HEADER in each
+ * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+ * If applicable, add the following below this CDDL HEADER, with the
+ * fields enclosed by brackets "[]" replaced with your own identifying
+ * information: Portions Copyright [yyyy] [name of copyright owner]
+ *
+ * CDDL HEADER END
+ */
+
+/*
+ * Copyright (c) 2018 by Delphix. All rights reserved.
+ */
+
+#include <sys/dataset_kstats.h>
+#include <sys/dmu_objset.h>
+#include <sys/dsl_dataset.h>
+#include <sys/spa.h>
+
+static dataset_kstat_values_t empty_dataset_kstats = {
+	{ "dataset_name",	KSTAT_DATA_STRING },
+	{ "writes",	KSTAT_DATA_UINT64 },
+	{ "nwritten",	KSTAT_DATA_UINT64 },
+	{ "reads",	KSTAT_DATA_UINT64 },
+	{ "nread",	KSTAT_DATA_UINT64 },
+};
+
+static int
+dataset_kstats_update(kstat_t *ksp, int rw)
+{
+	dataset_kstats_t *dk = ksp->ks_private;
+	ASSERT3P(dk->dk_kstats->ks_data, ==, ksp->ks_data);
+
+	if (rw == KSTAT_WRITE)
+		return (EACCES);
+
+	dataset_kstat_values_t *dkv = dk->dk_kstats->ks_data;
+	dkv->dkv_writes.value.ui64 =
+	    aggsum_value(&dk->dk_aggsums.das_writes);
+	dkv->dkv_nwritten.value.ui64 =
+	    aggsum_value(&dk->dk_aggsums.das_nwritten);
+	dkv->dkv_reads.value.ui64 =
+	    aggsum_value(&dk->dk_aggsums.das_reads);
+	dkv->dkv_nread.value.ui64 =
+	    aggsum_value(&dk->dk_aggsums.das_nread);
+
+	return (0);
+}
+
+void
+dataset_kstats_create(dataset_kstats_t *dk, objset_t *objset)
+{
+	/*
+	 * There should not be anything wrong with having kstats for
+	 * snapshots. Since we are not sure how useful they would be
+	 * though nor how much their memory overhead would matter in
+	 * a filesystem with many snapshots, we skip them for now.
+	 */
+	if (dmu_objset_is_snapshot(objset))
+		return;
+
+	/*
+	 * At the time of this writing, KSTAT_STRLEN is 255 in Linux,
+	 * and the spa_name can theoretically be up to 256 characters.
+	 * In reality though the spa_name can be 240 characters max
+	 * [see origin directory name check in pool_namecheck()]. Thus,
+	 * the naming scheme for the module name below should not cause
+	 * any truncations. In the event that a truncation does happen
+	 * though, due to some future change, we silently skip creating
+	 * the kstat and log the event.
+	 */
+	char kstat_module_name[KSTAT_STRLEN];
+	int n = snprintf(kstat_module_name, sizeof (kstat_module_name),
+	    "zfs/%s", spa_name(dmu_objset_spa(objset)));
+	if (n < 0) {
+		zfs_dbgmsg("failed to create dataset kstat for objset %lld: "
+		    " snprintf() for kstat module name returned %d",
+		    (unsigned long long)dmu_objset_id(objset), n);
+		return;
+	} else if (n >= KSTAT_STRLEN) {
+		zfs_dbgmsg("failed to create dataset kstat for objset %lld: "
+		    "kstat module name length (%d) exceeds limit (%d)",
+		    (unsigned long long)dmu_objset_id(objset),
+		    n, KSTAT_STRLEN);
+		return;
+	}
+
+	char kstat_name[KSTAT_STRLEN];
+	n = snprintf(kstat_name, sizeof (kstat_name), "objset-0x%llx",
+	    (unsigned long long)dmu_objset_id(objset));
+	if (n < 0) {
+		zfs_dbgmsg("failed to create dataset kstat for objset %lld: "
+		    " snprintf() for kstat name returned %d",
+		    (unsigned long long)dmu_objset_id(objset), n);
+		return;
+	}
+	ASSERT3U(n, <, KSTAT_STRLEN);
+
+	kstat_t *kstat = kstat_create(kstat_module_name, 0, kstat_name,
+	    "dataset", KSTAT_TYPE_NAMED,
+	    sizeof (empty_dataset_kstats) / sizeof (kstat_named_t),
+	    KSTAT_FLAG_VIRTUAL);
+	if (kstat == NULL)
+		return;
+
+	dataset_kstat_values_t *dk_kstats =
+	    kmem_alloc(sizeof (empty_dataset_kstats), KM_SLEEP);
+	bcopy(&empty_dataset_kstats, dk_kstats,
+	    sizeof (empty_dataset_kstats));
+
+	char *ds_name = kmem_zalloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
+	dsl_dataset_name(objset->os_dsl_dataset, ds_name);
+	KSTAT_NAMED_STR_PTR(&dk_kstats->dkv_ds_name) = ds_name;
+	KSTAT_NAMED_STR_BUFLEN(&dk_kstats->dkv_ds_name) =
+	    ZFS_MAX_DATASET_NAME_LEN;
+
+	kstat->ks_data = dk_kstats;
+	kstat->ks_update = dataset_kstats_update;
+	kstat->ks_private = dk;
+
+	kstat_install(kstat);
+	dk->dk_kstats = kstat;
+
+	aggsum_init(&dk->dk_aggsums.das_writes, 0);
+	aggsum_init(&dk->dk_aggsums.das_nwritten, 0);
+	aggsum_init(&dk->dk_aggsums.das_reads, 0);
+	aggsum_init(&dk->dk_aggsums.das_nread, 0);
+}
+
+void
+dataset_kstats_destroy(dataset_kstats_t *dk)
+{
+	if (dk->dk_kstats == NULL)
+		return;
+
+	dataset_kstat_values_t *dkv = dk->dk_kstats->ks_data;
+	kmem_free(KSTAT_NAMED_STR_PTR(&dkv->dkv_ds_name),
+	    KSTAT_NAMED_STR_BUFLEN(&dkv->dkv_ds_name));
+	kmem_free(dkv, sizeof (empty_dataset_kstats));
+
+	kstat_delete(dk->dk_kstats);
+	dk->dk_kstats = NULL;
+
+	aggsum_fini(&dk->dk_aggsums.das_writes);
+	aggsum_fini(&dk->dk_aggsums.das_nwritten);
+	aggsum_fini(&dk->dk_aggsums.das_reads);
+	aggsum_fini(&dk->dk_aggsums.das_nread);
+}
+
+void
+dataset_kstats_update_write_kstats(dataset_kstats_t *dk,
+    int64_t nwritten)
+{
+	ASSERT3S(nwritten, >=, 0);
+
+	if (dk->dk_kstats == NULL)
+		return;
+
+	aggsum_add(&dk->dk_aggsums.das_writes, 1);
+	aggsum_add(&dk->dk_aggsums.das_nwritten, nwritten);
+}
+
+void
+dataset_kstats_update_read_kstats(dataset_kstats_t *dk,
+    int64_t nread)
+{
+	ASSERT3S(nread, >=, 0);
+
+	if (dk->dk_kstats == NULL)
+		return;
+
+	aggsum_add(&dk->dk_aggsums.das_reads, 1);
+	aggsum_add(&dk->dk_aggsums.das_nread, nread);
+}