Multi-modifier protection (MMP)

Add multihost=on|off pool property to control MMP.  When enabled
a new thread writes uberblocks to the last slot in each label, at a
set frequency, to indicate to other hosts the pool is actively imported.
These uberblocks are the last synced uberblock with an updated
timestamp.  Property defaults to off.

During tryimport, find the "best" uberblock (newest txg and timestamp)
repeatedly, checking for change in the found uberblock.  Include the
results of the activity test in the config returned by tryimport.
These results are reported to user in "zpool import".

Allow the user to control the period between MMP writes, and the
duration of the activity test on import, via a new module parameter
zfs_multihost_interval.  The period is specified in milliseconds.  The
activity test duration is calculated from this value, and from the
mmp_delay in the "best" uberblock found initially.

Add a kstat interface to export statistics about Multiple Modifier
Protection (MMP) updates. Include the last synced txg number, the
timestamp, the delay since the last MMP update, the VDEV GUID, the VDEV
label that received the last MMP update, and the VDEV path.  Abbreviated
output below.

$ cat /proc/spl/kstat/zfs/mypool/multihost
31 0 0x01 10 880 105092382393521 105144180101111
txg   timestamp  mmp_delay   vdev_guid   vdev_label vdev_path
20468    261337  250274925   68396651780       3    /dev/sda
20468    261339  252023374   6267402363293     1    /dev/sdc
20468    261340  252000858   6698080955233     1    /dev/sdx
20468    261341  251980635   783892869810      2    /dev/sdy
20468    261342  253385953   8923255792467     3    /dev/sdd
20468    261344  253336622   042125143176      0    /dev/sdab
20468    261345  253310522   1200778101278     2    /dev/sde
20468    261346  253286429   0950576198362     2    /dev/sdt
20468    261347  253261545   96209817917       3    /dev/sds
20468    261349  253238188   8555725937673     3    /dev/sdb

Add a new tunable zfs_multihost_history to specify the number of MMP
updates to store history for. By default it is set to zero meaning that
no MMP statistics are stored.

When using ztest to generate activity, for automated tests of the MMP
function, some test functions interfere with the test.  For example, the
pool is exported to run zdb and then imported again.  Add a new ztest
function, "-M", to alter ztest behavior to prevent this.

Add new tests to verify the new functionality.  Tests provided by
Giuseppe Di Natale.

Reviewed by: Matthew Ahrens <[email protected]>
Reviewed-by: Giuseppe Di Natale <[email protected]>
Reviewed-by: Ned Bass <[email protected]>
Reviewed-by: Andreas Dilger <[email protected]>
Reviewed-by: Brian Behlendorf <[email protected]>
Signed-off-by: Olaf Faaland <[email protected]>
Closes #745
Closes #6279

10 files changed, 108 insertions, 2 deletions

diff --git a/include/sys/Makefile.am b/include/sys/Makefile.am
index 956643801..be606b8c6 100644
--- a/include/sys/Makefile.am
+++ b/include/sys/Makefile.am
@@ -36,6 +36,7 @@ COMMON_H = \
 	$(top_srcdir)/include/sys/efi_partition.h \
 	$(top_srcdir)/include/sys/metaslab.h \
 	$(top_srcdir)/include/sys/metaslab_impl.h \
+	$(top_srcdir)/include/sys/mmp.h \
 	$(top_srcdir)/include/sys/mntent.h \
 	$(top_srcdir)/include/sys/multilist.h \
 	$(top_srcdir)/include/sys/nvpair.h \
diff --git a/include/sys/dsl_pool.h b/include/sys/dsl_pool.h
index 674dd25b5..d2dabda6d 100644
--- a/include/sys/dsl_pool.h
+++ b/include/sys/dsl_pool.h
@@ -38,6 +38,7 @@
 #include <sys/bpobj.h>
 #include <sys/bptree.h>
 #include <sys/rrwlock.h>
+#include <sys/mmp.h>
 
 #ifdef	__cplusplus
 extern "C" {
diff --git a/include/sys/fs/zfs.h b/include/sys/fs/zfs.h
index 9fbcfbef3..13b25a695 100644
--- a/include/sys/fs/zfs.h
+++ b/include/sys/fs/zfs.h
@@ -223,6 +223,7 @@ typedef enum {
 	ZPOOL_PROP_MAXBLOCKSIZE,
 	ZPOOL_PROP_TNAME,
 	ZPOOL_PROP_MAXDNODESIZE,
+	ZPOOL_PROP_MULTIHOST,
 	ZPOOL_NUM_PROPS
 } zpool_prop_t;
 
@@ -651,6 +652,11 @@ typedef struct zpool_rewind_policy {
 #define	ZPOOL_CONFIG_VDEV_TOP_ZAP	"com.delphix:vdev_zap_top"
 #define	ZPOOL_CONFIG_VDEV_LEAF_ZAP	"com.delphix:vdev_zap_leaf"
 #define	ZPOOL_CONFIG_HAS_PER_VDEV_ZAPS	"com.delphix:has_per_vdev_zaps"
+#define	ZPOOL_CONFIG_MMP_STATE		"mmp_state"	/* not stored on disk */
+#define	ZPOOL_CONFIG_MMP_TXG		"mmp_txg"	/* not stored on disk */
+#define	ZPOOL_CONFIG_MMP_HOSTNAME	"mmp_hostname"	/* not stored on disk */
+#define	ZPOOL_CONFIG_MMP_HOSTID		"mmp_hostid"	/* not stored on disk */
+
 /*
  * The persistent vdev state is stored as separate values rather than a single
  * 'vdev_state' entry.  This is because a device can be in multiple states, such
@@ -744,7 +750,8 @@ typedef enum vdev_aux {
 	VDEV_AUX_EXTERNAL,	/* external diagnosis or forced fault	*/
 	VDEV_AUX_SPLIT_POOL,	/* vdev was split off into another pool	*/
 	VDEV_AUX_BAD_ASHIFT,	/* vdev ashift is invalid		*/
-	VDEV_AUX_EXTERNAL_PERSIST	/* persistent forced fault	*/
+	VDEV_AUX_EXTERNAL_PERSIST,	/* persistent forced fault	*/
+	VDEV_AUX_ACTIVE,	/* vdev active on a different host	*/
 } vdev_aux_t;
 
 /*
@@ -765,6 +772,16 @@ typedef enum pool_state {
 } pool_state_t;
 
 /*
+ * mmp state. The following states provide additional detail describing
+ * why a pool couldn't be safely imported.
+ */
+typedef enum mmp_state {
+	MMP_STATE_ACTIVE = 0,		/* In active use		*/
+	MMP_STATE_INACTIVE,		/* Inactive and safe to import	*/
+	MMP_STATE_NO_HOSTID		/* System hostid is not set	*/
+} mmp_state_t;
+
+/*
  * Scan Functions.
  */
 typedef enum pool_scan_func {
@@ -1126,6 +1143,7 @@ typedef enum {
 #define	ZFS_IMPORT_MISSING_LOG	0x4
 #define	ZFS_IMPORT_ONLY		0x8
 #define	ZFS_IMPORT_TEMP_NAME	0x10
+#define	ZFS_IMPORT_SKIP_MMP	0x20
 
 /*
  * Sysevent payload members.  ZFS will generate the following sysevents with the
diff --git a/include/sys/mmp.h b/include/sys/mmp.h
new file mode 100644
index 000000000..4da612d6a
--- /dev/null
+++ b/include/sys/mmp.h
@@ -0,0 +1,63 @@
+/*
+ * CDDL HEADER START
+ *
+ * This file and its contents are supplied under the terms of the
+ * Common Development and Distribution License ("CDDL"), version 1.0.
+ * You may only use this file in accordance with the terms of version
+ * 1.0 of the CDDL.
+ *
+ * A full copy of the text of the CDDL should have accompanied this
+ * source.  A copy of the CDDL is also available via the Internet at
+ * http://www.illumos.org/license/CDDL.
+ *
+ * CDDL HEADER END
+ */
+/*
+ * Copyright (C) 2017 by Lawrence Livermore National Security, LLC.
+ */
+
+#ifndef _SYS_MMP_H
+#define	_SYS_MMP_H
+
+#include <sys/spa.h>
+#include <sys/zfs_context.h>
+#include <sys/uberblock_impl.h>
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+#define	MMP_MIN_INTERVAL		100	/* ms */
+#define	MMP_DEFAULT_INTERVAL		1000	/* ms */
+#define	MMP_DEFAULT_IMPORT_INTERVALS	10
+#define	MMP_DEFAULT_FAIL_INTERVALS	5
+
+typedef struct mmp_thread {
+	kmutex_t	mmp_thread_lock; /* protect thread mgmt fields */
+	kcondvar_t	mmp_thread_cv;
+	kthread_t	*mmp_thread;
+	uint8_t		mmp_thread_exiting;
+	kmutex_t	mmp_io_lock;	/* protect below */
+	hrtime_t	mmp_last_write;	/* last successful MMP write */
+	uint64_t	mmp_delay;	/* decaying avg ns between MMP writes */
+	uberblock_t	mmp_ub;		/* last ub written by sync */
+	zio_t		*mmp_zio_root;	/* root of mmp write zios */
+} mmp_thread_t;
+
+
+extern void mmp_init(struct spa *spa);
+extern void mmp_fini(struct spa *spa);
+extern void mmp_thread_start(struct spa *spa);
+extern void mmp_thread_stop(struct spa *spa);
+extern void mmp_update_uberblock(struct spa *spa, struct uberblock *ub);
+
+/* Global tuning */
+extern ulong_t zfs_multihost_interval;
+extern uint_t zfs_multihost_fail_intervals;
+extern uint_t zfs_multihost_import_intervals;
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _SYS_MMP_H */
diff --git a/include/sys/spa.h b/include/sys/spa.h
index dd86aad40..de942ad2b 100644
--- a/include/sys/spa.h
+++ b/include/sys/spa.h
@@ -729,6 +729,7 @@ typedef struct spa_stats {
 	spa_stats_history_t	txg_history;
 	spa_stats_history_t	tx_assign_histogram;
 	spa_stats_history_t	io_history;
+	spa_stats_history_t	mmp_history;
 } spa_stats_t;
 
 typedef enum txg_state {
@@ -758,6 +759,8 @@ extern txg_stat_t *spa_txg_history_init_io(spa_t *, uint64_t,
     struct dsl_pool *);
 extern void spa_txg_history_fini_io(spa_t *, txg_stat_t *);
 extern void spa_tx_assign_add_nsecs(spa_t *spa, uint64_t nsecs);
+extern void spa_mmp_history_add(uint64_t txg, uint64_t timestamp,
+    uint64_t mmp_delay, vdev_t *vd, int label);
 
 /* Pool configuration locks */
 extern int spa_config_tryenter(spa_t *spa, int locks, void *tag, krw_t rw);
@@ -860,6 +863,8 @@ extern boolean_t spa_has_pending_synctask(spa_t *spa);
 extern int spa_maxblocksize(spa_t *spa);
 extern int spa_maxdnodesize(spa_t *spa);
 extern void zfs_blkptr_verify(spa_t *spa, const blkptr_t *bp);
+extern boolean_t spa_multihost(spa_t *spa);
+extern unsigned long spa_get_hostid(void);
 
 extern int spa_mode(spa_t *spa);
 extern uint64_t zfs_strtonum(const char *str, char **nptr);
diff --git a/include/sys/spa_impl.h b/include/sys/spa_impl.h
index 9142242cb..06de24421 100644
--- a/include/sys/spa_impl.h
+++ b/include/sys/spa_impl.h
@@ -275,6 +275,8 @@ struct spa {
 	spa_stats_t	spa_stats;		/* assorted spa statistics */
 	hrtime_t	spa_ccw_fail_time;	/* Conf cache write fail time */
 	taskq_t		*spa_zvol_taskq;	/* Taskq for minor management */
+	uint64_t	spa_multihost;		/* multihost aware (mmp) */
+	mmp_thread_t	spa_mmp;		/* multihost mmp thread */
 
 	/*
 	 * spa_refcount & spa_config_lock must be the last elements
diff --git a/include/sys/uberblock.h b/include/sys/uberblock.h
index 21e7ae0de..044e43838 100644
--- a/include/sys/uberblock.h
+++ b/include/sys/uberblock.h
@@ -40,7 +40,8 @@ extern "C" {
 typedef struct uberblock uberblock_t;
 
 extern int uberblock_verify(uberblock_t *);
-extern boolean_t uberblock_update(uberblock_t *, vdev_t *, uint64_t);
+extern boolean_t uberblock_update(uberblock_t *ub, vdev_t *rvd, uint64_t txg,
+    uint64_t mmp_delay);
 
 #ifdef	__cplusplus
 }
diff --git a/include/sys/uberblock_impl.h b/include/sys/uberblock_impl.h
index 6ab6aa313..9fdc70b91 100644
--- a/include/sys/uberblock_impl.h
+++ b/include/sys/uberblock_impl.h
@@ -43,6 +43,7 @@ extern "C" {
  */
 #define	UBERBLOCK_MAGIC		0x00bab10c		/* oo-ba-bloc!	*/
 #define	UBERBLOCK_SHIFT		10			/* up to 1K	*/
+#define	MMP_MAGIC		0xa11cea11		/* all-see-all  */
 
 struct uberblock {
 	uint64_t	ub_magic;	/* UBERBLOCK_MAGIC		*/
@@ -54,6 +55,11 @@ struct uberblock {
 
 	/* highest SPA_VERSION supported by software that wrote this txg */
 	uint64_t	ub_software_version;
+
+	/* Maybe missing in uberblocks we read, but always written */
+	uint64_t	ub_mmp_magic;	/* MMP_MAGIC			*/
+	uint64_t	ub_mmp_delay;	/* nanosec since last MMP write	*/
+	uint64_t	ub_mmp_seq;	/* reserved for sequence number	*/
 };
 
 #ifdef	__cplusplus
diff --git a/include/sys/vdev.h b/include/sys/vdev.h
index 63b4904c5..7157ef43f 100644
--- a/include/sys/vdev.h
+++ b/include/sys/vdev.h
@@ -155,6 +155,8 @@ extern int vdev_label_number(uint64_t psise, uint64_t offset);
 extern nvlist_t *vdev_label_read_config(vdev_t *vd, uint64_t txg);
 extern void vdev_uberblock_load(vdev_t *, struct uberblock *, nvlist_t **);
 extern void vdev_config_generate_stats(vdev_t *vd, nvlist_t *nv);
+extern void vdev_label_write(zio_t *zio, vdev_t *vd, int l, abd_t *buf, uint64_t
+    offset, uint64_t size, zio_done_func_t *done, void *private, int flags);
 
 typedef enum {
 	VDEV_LABEL_CREATE,	/* create/add a new device */
diff --git a/include/sys/vdev_impl.h b/include/sys/vdev_impl.h
index 835d2dbbf..7c5e54b08 100644
--- a/include/sys/vdev_impl.h
+++ b/include/sys/vdev_impl.h
@@ -238,6 +238,7 @@ struct vdev {
 	zio_t		*vdev_probe_zio; /* root of current probe	*/
 	vdev_aux_t	vdev_label_aux;	/* on-disk aux state		*/
 	uint64_t	vdev_leaf_zap;
+	hrtime_t	vdev_mmp_pending; /* 0 if write finished	*/
 
 	/*
 	 * For DTrace to work in userland (libzpool) context, these fields must
@@ -268,6 +269,12 @@ struct vdev {
 #define	VDEV_PHYS_SIZE		(112 << 10)
 #define	VDEV_UBERBLOCK_RING	(128 << 10)
 
+/*
+ * MMP blocks occupy the last MMP_BLOCKS_PER_LABEL slots in the uberblock
+ * ring when MMP is enabled.
+ */
+#define	MMP_BLOCKS_PER_LABEL	1
+
 /* The largest uberblock we support is 8k. */
 #define	MAX_UBERBLOCK_SHIFT (13)
 #define	VDEV_UBERBLOCK_SHIFT(vd)	\