Illumos #1051: zfs should handle imbalanced luns

Today zfs tries to allocate blocks evenly across all devices.
This means when devices are imbalanced zfs will use lots of
CPU searching for space on devices which tend to be pretty
full.  It should instead fail quickly on the full LUNs and
move onto devices which have more availability.

Reviewed by: Eric Schrock <[email protected]>
Reviewed by: Matt Ahrens <[email protected]>
Reviewed by: Adam Leventhal <[email protected]>
Reviewed by: Albert Lee <[email protected]>
Reviewed by: Gordon Ross <[email protected]>
Approved by: Garrett D'Amore <[email protected]>

References to Illumos issue and patch:
- https://www.illumos.org/issues/510
- https://github.com/illumos/illumos-gate/commit/5ead3ed965

Signed-off-by: Brian Behlendorf <[email protected]>
Issue #340

4 files changed, 15 insertions, 0 deletions

diff --git a/include/sys/metaslab.h b/include/sys/metaslab.h
index 583d6303b..2cf4d2b48 100644
--- a/include/sys/metaslab.h
+++ b/include/sys/metaslab.h
@@ -20,6 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2011 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_METASLAB_H
@@ -47,6 +48,8 @@ extern void metaslab_sync_reassess(metaslab_group_t *mg);
 #define	METASLAB_HINTBP_FAVOR	0x0
 #define	METASLAB_HINTBP_AVOID	0x1
 #define	METASLAB_GANG_HEADER	0x2
+#define	METASLAB_GANG_CHILD	0x4
+#define	METASLAB_GANG_AVOID	0x8
 
 extern int metaslab_alloc(spa_t *spa, metaslab_class_t *mc, uint64_t psize,
     blkptr_t *bp, int ncopies, uint64_t txg, blkptr_t *hintbp, int flags);
diff --git a/include/sys/metaslab_impl.h b/include/sys/metaslab_impl.h
index 07988dd51..6c670a162 100644
--- a/include/sys/metaslab_impl.h
+++ b/include/sys/metaslab_impl.h
@@ -21,6 +21,7 @@
 /*
  * Copyright 2009 Sun Microsystems, Inc.  All rights reserved.
  * Use is subject to license terms.
+ * Copyright (c) 2011 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_METASLAB_IMPL_H
@@ -52,6 +53,7 @@ struct metaslab_group {
 	avl_tree_t		mg_metaslab_tree;
 	uint64_t		mg_aliquot;
 	uint64_t		mg_bonus_area;
+	uint64_t		mg_alloc_failures;
 	int64_t			mg_bias;
 	int64_t			mg_activation_count;
 	metaslab_class_t	*mg_class;
diff --git a/include/sys/spa.h b/include/sys/spa.h
index 52737ebc2..c9028fb09 100644
--- a/include/sys/spa.h
+++ b/include/sys/spa.h
@@ -20,6 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2011 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_SPA_H
@@ -698,6 +699,13 @@ _NOTE(CONSTCOND) } while (0)
 #define	dprintf_bp(bp, fmt, ...)
 #endif
 
+extern boolean_t spa_debug_enabled(spa_t *spa);
+#define	spa_dbgmsg(spa, ...)			\
+{						\
+	if (spa_debug_enabled(spa))		\
+		zfs_dbgmsg(__VA_ARGS__);	\
+}
+
 extern int spa_mode_global;			/* mode, e.g. FREAD | FWRITE */
 
 #ifdef	__cplusplus
diff --git a/include/sys/spa_impl.h b/include/sys/spa_impl.h
index 1c34873b6..3f5cd9a73 100644
--- a/include/sys/spa_impl.h
+++ b/include/sys/spa_impl.h
@@ -20,6 +20,7 @@
  */
 /*
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2011 by Delphix. All rights reserved.
  */
 
 #ifndef _SYS_SPA_IMPL_H
@@ -196,6 +197,7 @@ struct spa {
 	kcondvar_t	spa_suspend_cv;		/* notification of resume */
 	uint8_t		spa_suspended;		/* pool is suspended */
 	uint8_t		spa_claiming;		/* pool is doing zil_claim() */
+	boolean_t	spa_debug;		/* debug enabled? */
 	boolean_t	spa_is_root;		/* pool is root */
 	int		spa_minref;		/* num refs when first opened */
 	int		spa_mode;		/* FREAD | FWRITE */