pool metadata spare

index cbccdfd..201ad4a 100644
@@ -60,6 +60,7 @@ static const struct flag _lv_flags[] = {
 	{LV_NOTSYNCED, "NOTSYNCED", STATUS_FLAG},
 	{LV_REBUILD, "REBUILD", STATUS_FLAG},
 	{LV_WRITEMOSTLY, "WRITEMOSTLY", STATUS_FLAG},
+	{POOL_METADATA_SPARE, NULL, 0},
 	{RAID, NULL, 0},
 	{RAID_META, NULL, 0},
 	{RAID_IMAGE, NULL, 0},

index 54a4a9c..59603bb 100644
@@ -617,6 +617,18 @@ static int _read_lvnames(struct format_instance *fid __attribute__((unused)),
 	if (timestamp && !lv_set_creation(lv, hostname, timestamp))
 		return_0;
 
+	if (!lv_is_visible(lv) && strstr(lv->name, "_pmspare")) {
+		if (vg->pool_metadata_spare_lv) {
+			log_error("Couldn't use another pool metadata spare "
+				  "logical volume %s/%s.", vg->name, lv->name);
+			return 0;
+		}
+		log_debug_metadata("Logical volume %s is pool metadata spare.",
+				   lv->name);
+		lv->status |= POOL_METADATA_SPARE;
+		vg->pool_metadata_spare_lv = lv;
+	}
+
 	return 1;
 }
 

index 412d29c..0250941 100644
@@ -99,6 +99,7 @@
 #define THIN_POOL_METADATA	UINT64_C(0x0000008000000000)	/* LV */
 
 #define LV_WRITEMOSTLY		UINT64_C(0x0000010000000000)	/* LV (RAID1) */
+#define POOL_METADATA_SPARE	UINT64_C(0x0000020000000000)	/* LV internal */
 
 /* Format features flags */
 #define FMT_SEGMENTS		0x00000001U	/* Arbitrary segment params? */

@@ -158,6 +159,7 @@
 #define lv_is_raid_type(lv)	(((lv)->status & (RAID | RAID_IMAGE | RAID_META)) ? 1 : 0)
 
 #define lv_is_virtual(lv)	(((lv)->status & VIRTUAL) ? 1 : 0)
+#define lv_is_pool_metadata_spare(lv)	(((lv)->status & POOL_METADATA_SPARE) ? 1 : 0)
 
 /* Ordered list - see lv_manip.c */
 typedef enum {

@@ -664,6 +666,8 @@ struct logical_volume *alloc_pool_metadata(struct logical_volume *pool_lv,
 					   uint32_t stripes, uint32_t stripe_size,
 					   uint64_t size, alloc_policy_t alloc,
 					   struct dm_list *pvh);
+int vg_set_pool_metadata_spare(struct logical_volume *lv);
+int vg_remove_pool_metadata_spare(struct volume_group *vg);
 
 /*
  * Activation options

index 96a636c..3796eb6 100644
@@ -2295,6 +2295,7 @@ int vg_validate(struct volume_group *vg)
 	unsigned hidden_lv_count = 0, lv_count = 0, lv_visible_count = 0;
 	unsigned pv_count = 0;
 	unsigned num_snapshots = 0;
+	unsigned spare_count = 0;
 	struct validate_hash vhash = { NULL };
 
 	if (vg->alloc == ALLOC_CLING_BY_TAGS) {

@@ -2474,6 +2475,19 @@ int vg_validate(struct volume_group *vg)
 			r = 0;
 		}
 
+		if (lv_is_pool_metadata_spare(lvl->lv)) {
+			if (++spare_count > 1) {
+				log_error(INTERNAL_ERROR "LV %s is %u. pool metadata spare (>1).",
+					  lvl->lv->name, spare_count);
+				r = 0;
+			}
+			if (vg->pool_metadata_spare_lv != lvl->lv) {
+				log_error(INTERNAL_ERROR "LV %s is not vg pool metadata spare.",
+					  lvl->lv->name);
+				r = 0;
+			}
+		}
+
 		if (!check_lv_segments(lvl->lv, 1)) {
 			log_error(INTERNAL_ERROR "LV segments corrupted in %s.",
 				  lvl->lv->name);

@@ -2526,6 +2540,13 @@ int vg_validate(struct volume_group *vg)
 		r = 0;
 	}
 
+	if (vg->pool_metadata_spare_lv &&
+	    !lv_is_pool_metadata_spare(vg->pool_metadata_spare_lv)) {
+		log_error(INTERNAL_ERROR "VG references non pool metadata spare LV %s.",
+			  vg->pool_metadata_spare_lv->name);
+		r = 0;
+	}
+
 	if (vg_max_lv_reached(vg))
 		stack;
 out:

74 lines not on this page · raw message

index 8db04d3..bf3a59e 100644
@@ -20,6 +20,7 @@ struct dm_pool;
 struct format_instance;
 struct dm_list;
 struct id;
+struct logical_volume;
 
 typedef enum {
 	ALLOC_INVALID,

@@ -121,6 +122,7 @@ struct volume_group {
 	uint32_t mda_copies; /* target number of mdas for this VG */
 
 	struct dm_hash_table *hostnames; /* map of creation hostnames */
+	struct logical_volume *pool_metadata_spare_lv; /* one per VG */
 };
 
 struct volume_group *alloc_vg(const char *pool_name, struct cmd_context *cmd,

index 9dcdb3d..8aa1f6c 100644
@@ -105,6 +105,7 @@ int apply_lvname_restrictions(const char *name)
 	static const char * const _reserved_strings[] = {
 		"_mlog",
 		"_mimage",
+		"_pmspare",
 		"_rimage",
 		"_rmeta",
 		"_vorigin",

index 0742562..31e0c83 100644
@@ -547,6 +547,7 @@ char *lv_attr_dup(struct dm_pool *mem, const struct logical_volume *lv)
 	else if (lv_is_origin(lv) || lv_is_external_origin(lv))
 		repstr[0] = (lv_is_merging_origin(lv)) ? 'O' : 'o';
 	else if (lv_is_thin_pool_metadata(lv) ||
+		 lv_is_pool_metadata_spare(lv) ||
 		 (lv->status & RAID_META))
 		repstr[0] = 'e';
 	else if (lv->status & RAID)

index 9d67e71..d86d070 100644
@@ -4521,7 +4521,9 @@ int lv_remove_with_dependencies(struct cmd_context *cmd, struct logical_volume *
 {
 	percent_t snap_percent;
 	struct dm_list *snh, *snht;
+	struct lv_list *lvl;
 	struct lvinfo info;
+	int found;
 
 	if (lv_is_cow(lv)) {
 		/*

@@ -4579,6 +4581,30 @@ int lv_remove_with_dependencies(struct cmd_context *cmd, struct logical_volume *
 	    !_lv_remove_segs_using_this_lv(cmd, lv, force, level, "pool"))
 		return_0;
 
+	if (lv_is_thin_pool(lv) && lv->vg->pool_metadata_spare_lv) {
+		found = 0;
+		/* When removing last thin pool, remove spare as well */
+		dm_list_iterate_items(lvl, &lv->vg->lvs)
+			if (lv_is_thin_pool(lvl->lv) && lvl->lv != lv) {
+				found = 1;
+				break;
+			}
+
+		if (!found &&
+		    !lv_remove_with_dependencies(cmd, lv->vg->pool_metadata_spare_lv,
+						 force, level + 1))
+			return_0;
+	}
+
+	if (lv_is_pool_metadata_spare(lv) &&
+	    (force == PROMPT) &&
+	    (yes_no_prompt("Removal of pool metadata spare logical volume \"%s\" "
+			   "disables automatic recovery attempts after damage "
+			   "to a thin pool. Proceed? [y/n]: ", lv->name) == 'n')) {
+		log_error("Logical volume \"%s\" not removed.", lv->name);
+		return 0;
+	}
+
 	return lv_remove_single(cmd, lv, force);
 }
 

index 73312ca..2768640 100644
@@ -74,6 +74,7 @@ arg(originname_ARG, '\0', "originname", string_arg, 0)
 arg(poll_ARG, '\0', "poll", yes_no_arg, 0)
 arg(poolmetadata_ARG, '\0', "poolmetadata", string_arg, 0)
 arg(poolmetadatasize_ARG, '\0', "poolmetadatasize", size_mb_arg, 0)
+arg(poolmetadataspare_ARG, '\0', "poolmetadataspare", yes_no_arg, 0)
 arg(discards_ARG, '\0', "discards", discards_arg, 0)
 arg(force_long_ARG, '\0', "force", NULL, ARG_COUNTABLE)
 arg(stripes_long_ARG, '\0', "stripes", int_arg, 0)

index f84aa35..89d8764 100644
@@ -93,6 +93,7 @@ xx(lvchange,
    "\t[-r|--readahead ReadAheadSectors|auto|none]\n"
    "\t[--refresh]\n"
    "\t[--resync]\n"
+   "\t[--poolmetadataspare {y|n}]\n"
    "\t[--syncaction {check|repair}\n"
    "\t[--sysinit]\n"
    "\t[-t|--test]\n"

@@ -108,7 +109,8 @@ xx(lvchange,
    discards_ARG, detachprofile_ARG, force_ARG, ignorelockingfailure_ARG,
    ignoremonitoring_ARG, major_ARG, minor_ARG, monitor_ARG, minrecoveryrate_ARG,
    maxrecoveryrate_ARG, noudevsync_ARG, partial_ARG, permission_ARG,
-   persistent_ARG, poll_ARG, readahead_ARG, resync_ARG, refresh_ARG, addtag_ARG,
+   persistent_ARG, poll_ARG, poolmetadataspare_ARG,
+   readahead_ARG, resync_ARG, refresh_ARG, addtag_ARG,
    deltag_ARG, syncaction_ARG, sysinit_ARG, test_ARG, writebehind_ARG,
    writemostly_ARG, zero_ARG)
 

103 lines not on this page · raw message

index 7ee89ec..3cd92a6 100644
@@ -73,6 +73,7 @@
 #define DEFAULT_THIN_POOL_CHUNK_SIZE	    64	  /* KB */
 #define DEFAULT_THIN_POOL_DISCARDS "passdown"
 #define DEFAULT_THIN_POOL_ZERO 1
+#define DEFAULT_POOL_METADATA_SPARE 1
 
 #define DEFAULT_UMASK 0077
 

index 0250941..de69fd7 100644
@@ -668,6 +668,8 @@ struct logical_volume *alloc_pool_metadata(struct logical_volume *pool_lv,
 					   struct dm_list *pvh);
 int vg_set_pool_metadata_spare(struct logical_volume *lv);
 int vg_remove_pool_metadata_spare(struct volume_group *vg);
+int handle_pool_metadata_spare(struct volume_group *vg, uint32_t extents,
+			       struct dm_list *pvh, int poolmetadataspare);
 
 /*
  * Activation options

@@ -692,6 +694,7 @@ struct lvcreate_params {
 	int minor; /* all */
 	int log_count; /* mirror */
 	int nosync; /* mirror */
+	int poolmetadataspare; /* thin pool */
 	activation_change_t activate; /* non-snapshot, non-mirror */
 	thin_discards_t discards;     /* thin */
 

104 lines not on this page · raw message

index 89d8764..3aec4bc 100644
@@ -210,6 +210,7 @@ xx(lvcreate,
    "\t[-T|--thin  [-c|--chunksize  ChunkSize]\n"
    "\t  [--discards {ignore|nopassdown|passdown}]\n"
    "\t  [--poolmetadatasize MetadataSize[bBsSkKmMgG]]]\n"
+   "\t  [--poolmetadataspare {y|n}]\n"
    "\t[--thinpool ThinPoolLogicalVolume{Name|Path}]\n"
    "\t[-t|--test]\n"
    "\t[--type VolumeType]\n"

@@ -253,10 +254,11 @@ xx(lvcreate,
    chunksize_ARG, contiguous_ARG, corelog_ARG, discards_ARG, extents_ARG,
    ignoremonitoring_ARG, major_ARG, minor_ARG, mirrorlog_ARG, mirrors_ARG,
    monitor_ARG, minrecoveryrate_ARG, maxrecoveryrate_ARG, name_ARG, nosync_ARG,
-   noudevsync_ARG, permission_ARG,
-   persistent_ARG, readahead_ARG, regionsize_ARG, size_ARG, snapshot_ARG,
-   stripes_ARG, stripesize_ARG, test_ARG, thin_ARG, thinpool_ARG, type_ARG,
-   virtualoriginsize_ARG, poolmetadatasize_ARG, virtualsize_ARG, zero_ARG)
+   noudevsync_ARG, permission_ARG, persistent_ARG, poolmetadatasize_ARG,
+   poolmetadataspare_ARG, readahead_ARG, regionsize_ARG,
+   size_ARG, snapshot_ARG, stripes_ARG, stripesize_ARG,
+   test_ARG, thin_ARG, thinpool_ARG, type_ARG,
+   virtualoriginsize_ARG, virtualsize_ARG, zero_ARG)
 
 xx(lvdisplay,
    "Display information about a logical volume",

index 698d72f..caecc46 100644
@@ -1,6 +1,6 @@
 /*
  * Copyright (C) 2001-2004 Sistina Software, Inc. All rights reserved.
- * Copyright (C) 2004-2012 Red Hat, Inc. All rights reserved.
+ * Copyright (C) 2004-2013 Red Hat, Inc. All rights reserved.
  *
  * This file is part of LVM2.
  *

@@ -887,6 +887,14 @@ static int _lvcreate_params(struct lvcreate_params *lp,
 		return 0;
 	}
 
+	if (lp->create_thin_pool) {
+		/* TODO: add lvm.conf default y|n */
+		lp->poolmetadataspare = arg_int_value(cmd, poolmetadataspare_ARG,
+						      DEFAULT_POOL_METADATA_SPARE);
+	} else if (arg_count(cmd, poolmetadataspare_ARG)) {
+		log_error("--poolmetadataspare is only available with thin pool creation.");
+		return 0;
+	}
 	/*
 	 * Allocation parameters
 	 */

@@ -945,6 +953,7 @@ static int _check_thin_parameters(struct volume_group *vg, struct lvcreate_param
 			contiguous_ARG,
 			discards_ARG,
 			poolmetadatasize_ARG,
+			poolmetadataspare_ARG,
 			stripes_ARG,
 			stripesize_ARG,
 			zero_ARG

@@ -1079,9 +1088,14 @@ int lvcreate(struct cmd_context *cmd, int argc, char **argv)
 	if (seg_is_thin(&lp) && !_validate_internal_thin_processing(&lp))
 		goto_out;
 
-	if (lp.create_thin_pool)
+	if (lp.create_thin_pool) {
+		if (!handle_pool_metadata_spare(vg, lp.poolmetadataextents,
+						lp.pvh, lp.poolmetadataspare))
+			goto_out;
+
 		log_verbose("Making thin pool %s in VG %s using segtype %s",
 			    lp.pool ? : "with generated name", lp.vg_name, lp.segtype->name);
+	}
 
 	if (lp.thin)
 		log_verbose("Making thin LV %s in pool %s in VG %s%s%s using segtype %s",

index 3aec4bc..cb4ef44 100644
@@ -168,6 +168,7 @@ xx(lvconvert,
    "\t[--poolmetadata ThinMetadataLogicalVolume[Path] |\n"
    "\t [--poolmetadatasize size]\n"
    "\t [-r|--readahead ReadAheadSectors|auto|none]\n"
+   "\t [--poolmetadataspare {y|n}]\n"
    "\t [--stripes Stripes [-I|--stripesize StripeSize]]]\n"
    "\t[-T|--thin ExternalLogicalVolume[Path]\n"
    "\t [--originname NewExternalOriginVolumeName]]\n"

@@ -178,8 +179,9 @@ xx(lvconvert,
    merge_ARG, mirrorlog_ARG, mirrors_ARG, name_ARG, noudevsync_ARG,
    readahead_ARG, regionsize_ARG, repair_ARG, replace_ARG, snapshot_ARG, splitmirrors_ARG,
    trackchanges_ARG, type_ARG, stripes_long_ARG, stripesize_ARG, test_ARG,
-   chunksize_ARG, discards_ARG, poolmetadata_ARG, poolmetadatasize_ARG,
-   originname_ARG, thin_ARG, thinpool_ARG,
+   chunksize_ARG, discards_ARG, originname_ARG,
+   poolmetadata_ARG, poolmetadatasize_ARG, poolmetadataspare_ARG,
+   thin_ARG, thinpool_ARG,
    use_policies_ARG, force_ARG, zero_ARG)
 
 xx(lvcreate,

index 3ea1987..7cbf3d4 100644
@@ -22,6 +22,7 @@ struct lvconvert_params {
 	int snapshot;
 	int merge;
 	int merge_mirror;
+	int poolmetadataspare;
 	int thin;
 	int yes;
 	int zero;

@@ -254,6 +255,9 @@ static int _read_params(struct lvconvert_params *lp, struct cmd_context *cmd,
 	} else if (arg_count(cmd, discards_ARG)) {
 		log_error("--discards is only valid with --thinpool.");
 		return 0;
+	} else if (arg_count(cmd, poolmetadataspare_ARG)) {
+		log_error("--poolmetadataspare is only valid with --thinpool.");
+		return 0;
 	}
 
 	/*

@@ -487,6 +491,9 @@ static int _read_params(struct lvconvert_params *lp, struct cmd_context *cmd,
 			return_0;
 	}
 
+	/* TODO: default in lvm.conf ? */
+	lp->poolmetadataspare = arg_int_value(cmd, poolmetadataspare_ARG,
+					      DEFAULT_POOL_METADATA_SPARE);
 	lp->force = arg_count(cmd, force_ARG);
 	lp->yes = arg_count(cmd, yes_ARG);
 

@@ -2159,6 +2166,10 @@ static int _lvconvert_thinpool(struct cmd_context *cmd,
 		return 0;
 	}
 
+	if (!handle_pool_metadata_spare(pool_lv->vg, metadata_lv->le_count,
+					lp->pvh, lp->poolmetadataspare))
+		return_0;
+
 	old_name = data_lv->name; /* Use for pool name */
 	/*
 	 * Since we wish to have underlaying devs to match _tdata

index cb4ef44..1ce32eb 100644
@@ -784,6 +784,7 @@ xx(vgchange,
    "\t[--monitor {y|n}]\n"
    "\t[--[vg]metadatacopies #copies] " "\n"
    "\t[--poll {y|n}]\n"
+   "\t[--poolmetadataspare {y|n}]\n"
    "\t[--noudevsync]\n"
    "\t[--refresh]\n"
    "\t[--sysinit]\n"

@@ -806,6 +807,7 @@ xx(vgchange,
    ignorelockingfailure_ARG, ignoremonitoring_ARG, logicalvolume_ARG,
    maxphysicalvolumes_ARG, monitor_ARG, noudevsync_ARG, metadatacopies_ARG,
    vgmetadatacopies_ARG, partial_ARG, physicalextentsize_ARG, poll_ARG,
+   poolmetadataspare_ARG,
    refresh_ARG, resizeable_ARG, resizable_ARG, sysinit_ARG, test_ARG, uuid_ARG)
 
 xx(vgck,

index d3e93f2..af75857 100644
@@ -443,6 +443,25 @@ static int _vgchange_profile(struct cmd_context *cmd,
 	return 1;
 }
 
+static int _vgchange_pool_metadata_spare(struct cmd_context *cmd,
+					 struct volume_group *vg)
+{
+	int pms = arg_int_value(cmd, poolmetadataspare_ARG, 0);
+	/* Defaults not used here, value is given on cmdline */
+
+	if (pms) {
+		if (!handle_pool_metadata_spare(vg, 0, NULL, 1))
+			return_0;
+	} else if (vg->pool_metadata_spare_lv) {
+		if (!vg_remove_pool_metadata_spare(vg))
+			return_0;
+	} else
+		log_warn("Volume group \"%s\" is already without "
+			 "pool metadata spare volume.", vg->name);
+
+	return 1;
+}
+
 static int vgchange_single(struct cmd_context *cmd, const char *vg_name,
 			   struct volume_group *vg,
 			   void *handle __attribute__((unused)))

@@ -464,6 +483,7 @@ static int vgchange_single(struct cmd_context *cmd, const char *vg_name,
 		{ clustered_ARG, &_vgchange_clustered },
 		{ vgmetadatacopies_ARG, &_vgchange_metadata_copies },
 		{ profile_ARG, &_vgchange_profile},
+		{ poolmetadataspare_ARG, &_vgchange_pool_metadata_spare},
 		{ detachprofile_ARG, &_vgchange_profile},
 	};
 

@@ -545,6 +565,7 @@ int vgchange(struct cmd_context *cmd, int argc, char **argv)
 		arg_count(cmd, resizeable_ARG) ||
 		arg_count(cmd, uuid_ARG) ||
 		arg_count(cmd, physicalextentsize_ARG) ||
+		arg_count(cmd, poolmetadataspare_ARG) ||
 		arg_count(cmd, clustered_ARG) ||
 		arg_count(cmd, alloc_ARG) ||
 		arg_count(cmd, vgmetadatacopies_ARG);

@@ -557,8 +578,8 @@ int vgchange(struct cmd_context *cmd, int argc, char **argv)
 	    !arg_count(cmd, refresh_ARG)) {
 		log_error("Need 1 or more of -a, -c, -l, -p, -s, -x, "
 			  "--refresh, --uuid, --alloc, --addtag, --deltag, "
-			  "--monitor, --poll, --vgmetadatacopies or "
-			  "--metadatacopies");
+			  "--monitor, --poll, --poolmetadataspare "
+			  "--vgmetadatacopies or --metadatacopies");
 		return EINVALID_CMD_LINE;
 	}