1metaset(1M) System Administration Commands metaset(1M)
2
6 metaset - configure disk sets
7
9 /usr/sbin/metaset -s setname [-M-a -h hostname]
10 /usr/sbin/metaset -s setname -A {enable | disable}
13 /usr/sbin/metaset -s setname [-A {enable | disable}] -a -h hostname...
16 /usr/sbin/metaset -s setname -a [-l length] [-L] drivename...
19 /usr/sbin/metaset -s setname -C {take | release | purge}
22 /usr/sbin/metaset -s setname -d [-f] -h hostname...
25 /usr/sbin/metaset -s setname -d [-f] drivename...
28 /usr/sbin/metaset -s setname -j
31 /usr/sbin/metaset -s setname -r
34 /usr/sbin/metaset -s setname -w
37 /usr/sbin/metaset -s setname -t [-f] [-u tagnumber] [y]
40 /usr/sbin/metaset -s setname -b
43 /usr/sbin/metaset -s setname -P
46 /usr/sbin/metaset -s setname -q
49 /usr/sbin/metaset -s setname -o [-h hostname]
52 /usr/sbin/metaset [-s setname]
55 /usr/sbin/metaset [-s setname] -a | -d
58 [ [m] mediator_host_list]
59
62 The metaset command administers sets of disks in named disk sets. Named
63 disk sets include any disk set that is not in the local set. While disk
64 sets enable a high-availability configuration, Solaris Volume Manager
65 itself does not actually provide a high-availability environment.
66 A single-owner disk set configuration manages storage on a SAN or fab‐
69 ric-attached storage, or provides namespace control and state database
70 replica management for a specified set of disks.
71 In a shared disk set configuration, multiple hosts are physically con‐
74 nected to the same set of disks. When one host fails, another host has
75 exclusive access to the disks. Each host can control a shared disk set,
76 but only one host can control it at a time.
77 When you add a new disk to any disk set, Solaris Volume Manager checks
80 the disk format. If necessary, it repartitions the disk to ensure that
81 the disk has an appropriately configured reserved slice 7 (or slice 6
82 on an EFI labelled device) with adequate space for a state database
83 replica. The precise size of slice 7 (or slice 6 on an EFI labelled
84 device) depends on the disk geometry. For tradtional disk sets, the
85 slice is no less than 4 Mbytes, and probably closer to 6 Mbytes,
86 depending on where the cylinder boundaries lie. For multi-owner disk
87 sets, the slice is a minimum of 256 Mbytes. The minimal size for slice
88 7 might change in the future. This change is based on a variety of fac‐
89 tors, including the size of the state database replica and information
90 to be stored in the state database replica.
91 For use in disk sets, disks must have a dedicated slice (six or seven)
94 that meets specific criteria:
95 o The slice must start at sector 0
97 o The slice must include enough space for disk label
99 o The state database replicas cannot be mounted
101 o The slice does not overlap with any other slices, including
103 slice 2
104 If the existing partition table does not meet these criteria, or if the
107 -L flag is specified, Solaris Volume Manager repartitions the disk. A
108 small portion of each drive is reserved in slice 7 (or slice 6 on an
109 EFI labelled device) for use by Solaris Volume Manager. The remainder
110 of the space on each drive is placed into slice 0. Any existing data on
111 the disks is lost by repartitioning.
112 After you add a drive to a disk set, it can be repartitioned as neces‐
115 sary, with the exception that slice 7 (or slice 6 on an EFI labelled
116 device) is not altered in any way.
117 After a disk set is created and metadevices are set up within the set,
120 the metadevice name is in the following form:
121 /dev/md/setname/{dsk,rdsk}/dnumber
124 where setname is the name of the disk set, and number is the number of
127 the metadevice (0-127).
128 If you have disk sets that you upgraded from Solstice DiskSuite soft‐
131 ware, the default state database replica size on those sets is 1034
132 blocks, not the 8192 block size from Solaris Volume Manager. Also,
133 slice 7 on the disks that were added under Solstice DiskSuite are cor‐
134 respondingly smaller than slice 7 on disks that were added under
135 Solaris Volume Manager.
136 If disks you add to a disk set have acceptable slice 7s (that start at
139 cylinder 0 and that have sufficient space for the state database
140 replica), they are not reformatted.
141 Hot spare pools within local disk sets use standard Solaris Volume Man‐
144 ager naming conventions. Hot spare pools with shared disk sets use the
145 following convention:
146 setname/hot_spare_pool
149 where setname is the name of the disk set, and hot_spare_pool is the
152 name of the hot spare pool associated with the disk set.
153 Multi-node Environment
155 To create and work with a disk set in a multi—node environment, root
156 must be a member of Group 14 on all hosts, or the /.rhosts file must
157 contain an entry for all other host names. This is not required in a
158 SunCluster 3.x enviroment.
159 Tagged data
161 Tagged data occurs when there are different versions of a disk set's
162 replicas. This tagged data consists of the set owner's nodename, the
163 hardware serial number of the owner and the time it was written out to
164 the available replicas. The system administer can use this information
165 to determine which replica contains the correct data.
166 When a disk set is configured with an even number of storage enclosures
169 and has replicas balanced across them evenly, it is possible that up to
170 half of the replicas can be lost (for example, through a power failure
171 of half of the storage enclosures). After the enclosure that went down
172 is rebooted, half of the replicas are not recognized by SVM. When the
173 set is retaken, the metaset command returns an error of "stale data‐
174 bases", and all of the metadevices are in a read-only state.
175 Some of the replicas that are not recognized need to be deleted. The
178 action of deleting the replicas also causes updates to the replicas
179 that are not being deleted. In a dual hosted disk set environment, the
180 second node can access the deleted replicas instead of the existing
181 replicas when it takes the set. This leads to the possibility of get‐
182 ting the wrong replica record on a disk set take. An error message is
183 displayed, and user intervention is required.
184 Use the -q to query the disk set and the -t, -u, and -y, options to
187 select the tag and take the disk set. See OPTIONS.
188 Mediator Configuration
190 SVM provides support for a low-end HA solution consisting of two hosts
191 that share only two strings of drives. The hosts in this type of con‐
192 figuration, referred to as mediators or mediator hosts, run a special
193 daemon, rpc.metamedd(1M). The mediator hosts take on additional respon‐
194 sibilities to ensure that data is available in the case of host or
195 drive failures.
196 A mediator configuration can survive the failure of a single host or a
199 single string of drives, without administrative intervention. If both a
200 host and a string of drives fail (multiple failures), the integrity of
201 the data cannot be guaranteed. At this point, administrative interven‐
202 tion is required to make the data accessible. See mediator(7D) for fur‐
203 ther details.
204 Use the -m option to add or delete a mediator host. See OPTIONS.
207
209 The following options are supported:
210 -a drivename
212 Add drives or hosts to the named set. For a drive to be accepted
214 into a set, the drive must not be in use within another metadevice
215 or disk set, mounted on, or swapped on. When the drive is accepted
216 into the set, it is repartitioned and the metadevice state database
217 replica (for the set) can be placed on it. However, if a slice 7
218 (or slice 6 on an EFI labelled device), starts at cylinder 0, and
219 is large enough to hold a state database replica, then the disk is
220 not repartitioned. Also, a drive is not accepted if it cannot be
221 found on all hosts specified as part of the set. This means that if
222 a host within the specified set is unreachable due to network prob‐
223 lems, or is administratively down, the add fails.
224 Specify a drive name in the form cnumtnumdnum. Do not specify a
226 slice number (snum). For drives in a Sun Cluster, you must specify
227 a complete pathname for each drive. Such a name has the form:
228 /dev/did/[r]dsk/dnum
230 -a | -d | -m mediator_host_list
235 Add (-a) or delete (-d) mediator hosts to the specified disk set. A
237 mediator_host_list is the nodename(4) of the mediator host to be
238 added and (for adding) up to two other aliases for the mediator
239 host. The nodename and aliases for each mediator host are separated
240 only by commas. Up to three mediator hosts can be specified for the
241 named disk set. Specify only the nodename of that host as the argu‐
242 ment to -m to delete a mediator host.
243 In a single metaset command you can add or delete up to three medi‐
245 ator hosts. See EXAMPLES.
246 -A {enable | disable}
249 Specify auto-take status for a disk set. If auto-take is enabled
251 for a set, the disk set is automatically taken at boot, and file
252 systems on volumes within the disk set can be mounted through
253 /etc/vfstab entries. Only a single host can be associated with an
254 auto-take set, so attempts to add a second host to an auto-take set
255 or attempts to configure a disk set with multiple hosts as auto-
256 take fails with an error message. Disabling auto-take status for a
257 specific disk set causes the disk set to revert to normal behavior.
258 That is, the disk set is potentially shared (non-concurrently)
259 among hosts, and unavailable for mounting through /etc/vfstab.
260 -b
263 Insure that the replicas are distributed according to the replica
265 layout algorithm. This can be invoked at any time, and does nothing
266 if the replicas are correctly distributed. In cases where the user
267 has used the metadb command to manually remove or add replicas,
268 this command can be used to insure that the distribution of repli‐
269 cas matches the replica layout algorithm.
270 -C {take | release | purge}
273 Do not interact with the Cluster Framework when used in a Sun Clus‐
275 ter 3 environment. In effect, this means do not modify the Cluster
276 Configuration Repository. These options should only be used to fix
277 a broken disk set configuration.
278 take
280 Take ownership of the disk set but do not inform the Cluster
282 Framework that the disk set is available. This option is not
283 for use with a multi-owner disk set.
284 release
287 Release ownership of the disk set without informing the Cluster
289 Framework. This option should only be used if the disk set own‐
290 ership was taken with the corresponding -C take option. This
291 option is not for use with a multi-owner disk set.
292 purge
295 Remove the disk set without informing the Cluster Framework
297 that the disk set has been purged. This option should only be
298 used when the disk set is not accessible and requires rebuild‐
299 ing.
300 -d drivename
304 Delete drives or hosts from the named disk set. For a drive to be
306 deleted, it must not be in use within the set. The last host cannot
307 be deleted unless all of the drives within the set are deleted.
308 Deleting the last host in a disk set destroys the disk set.
309 Specify a drive name in the form cnumtnumdnum. Do not specify a
311 slice number (snum). For drives in a Sun Cluster, you must specify
312 a complete pathname for each drive. Such a name has the form:
313 /dev/did/[r]dsk/dnum
315 This option fails on a multi-owner disk set if attempting to with‐
318 draw the master node while other nodes are in the set.
319 -f
322 Force one of three actions to occur: takes ownership of a disk set
324 when used with -t; deletes the last disk drive from the disk set;
325 or deletes the last host from the disk set. Deleting the last drive
326 or host from a disk set requires the -d option.
327 When used to forcibly take ownership of the disk set, this causes
329 the disk set to be grabbed whether or not another host owns the
330 set. All of the disks within the set are taken over (reserved) and
331 fail fast is enabled, causing the other host to panic if it had
332 disk set ownership. The metadevice state database is read in by the
333 host performing the take, and the shared metadevices contained in
334 the set are accessible.
335 You can use this option to delete the last drive in the disk set,
337 because this drive would implicitly contain the last state database
338 replica.
339 You can use -f option to delete hosts from a set. When specified
341 with a partial list of hosts, it can be used for one-host adminis‐
342 tration. One-host administration could be useful when a host is
343 known to be non-functional, thus avoiding timeouts and failed com‐
344 mands. When specified with a complete list of hosts, the set is
345 completely deleted. It is generally specified with a complete list
346 of hosts to clean up after one-host administration has been per‐
347 formed.
348 -h hostname...
351 Specify one or more host names to be added to or deleted from a
353 disk set. Adding the first host creates the set. The last host can‐
354 not be deleted unless all of the drives within the set have been
355 deleted. The host name is not accepted if all of the drives within
356 the set cannot be found on the specified host. The host name is the
357 same name found in /etc/nodename.
358 -j
361 Join a host to the owner list for a multi-owner disk set. The con‐
363 cepts of take and release, used with traditional disk sets, do not
364 apply to multi-owner sets, because multiple owners are allowed.
365 As a host boots and is brought online, it must go through three
367 configuration levels to be able to use a multi-owner disk set:
368 1. It must be included in the cluster nodelist, which hap‐
370 pens automatically in a cluster or single-node sitatu‐
371 ion.
372 2. It must be added to the multi-owner disk set with the -a
374 -h options documented elsewhere in this man page
375 3. It must join the set. When the host is first added to
377 the set, it is automatically joined.
378 On manual restarts, the administrator must manually issue
379 metaset -s multinodesetname -j
381 to join the host to the owner list. After the cluster reconfigura‐
384 tion, when the host reenters the cluster, the node is automatically
385 joined to the set. The metaset -j command joins the host to all
386 multi-owner sets that the host has been added to. In a single node
387 situation, joining the node to the disk set starts any necessary
388 resynchronizations.
389 -L
392 When adding a disk to a disk set, force the disk to be reparti‐
394 tioned using the standard Solaris Volume Manager algorithm. See
395 DESCRIPTION.
396 -l length
399 Set the size (in blocks) for the metadevice state database replica.
401 The length can only be set when adding a new drive; it cannot be
402 changed on an existing drive. The default (and maximum) size is
403 8192 blocks, which should be appropriate for most configurations.
404 Replica sizes of less than 128 blocks are not recommended.
405 -M
408 Specify that the disk set to be created or modified is a multi-
410 owner disk set that supports multiple concurrent owners.
411 This option is required when creating a multi-owner disk set. Its
413 use is optional on all other operations on a multi-owner disk set
414 and has no effect. Existing disk sets cannot be converted to multi-
415 owner sets.
416 -o
419 Return an exit status of 0 if the local host or the host specified
421 with the -h option is the owner of the disk set.
422 -P
425 Purge the named disk set from the node on which the metaset command
427 is run. The disk set must not be owned by the node that runs this
428 command. If the node does own the disk set, the command fails.
429 If you need to delete a disk set but cannot take ownership of the
431 set, use the -P option.
432 -q
435 Displays an enumerated list of tags pertaining to ``tagged data''
437 that can be encountered during a take of the ownership of a disk
438 set.
439 This option is not for use with a multi-owner disk set.
441 -r
444 Release ownership of a disk set. All of the disks within the set
446 are released. The metadevices set up within the set are no longer
447 accessible.
448 This option is not for use with a multi-owner disk set.
450 -s setname
453 Specify the name of a disk set on which metaset works. If no set‐
455 name is specified, all disk sets are returned.
456 -t
459 Take ownership of a disk set safely. If metaset finds that another
461 host owns the set, this host is not be allowed to take ownership of
462 the set. If the set is not owned by any other host, all the disks
463 within the set are owned by the host on which metaset was executed.
464 The metadevice state database is read in, and the shared metade‐
465 vices contained in the set become accessible. The -t option takes a
466 disk set that has stale databases. When the databases are stale,
467 metaset exits with code 66, and prints a message. At that point,
468 the only operations permitted are the addition and deletion of
469 replicas. Once the addition or deletion of the replicas has been
470 completed, the disk set should be released and retaken to gain full
471 access to the data.
472 This option is not for use with a multi-owner disk set.
474 -u tagnumber
477 Once a tag has been selected, a subsequent take with -u tagnumber
479 can be executed to select the data associated with the given tag‐
480 number.
481 w
484 Withdraws a host from the owner list for a multi-owner disk set.
486 The concepts of take and release, used with traditional disk sets,
487 do not apply to multi-owner sets, because multiple owners are
488 allowed.
489 Instead of releasing a set, a host can issue
491 metaset -s multinodesetname -w
493 to withdraw from the owner list. A host automatically withdraws on
496 a reboot, but can be manually withdrawn if it should not be able to
497 use the set, but should be able to rejoin at a later time. A host
498 that withdrew due to a reboot can still appear joined from other
499 hosts in the set until a reconfiguration cycle occurs.
500 metaset -w withdraws from ownership of all multi-owner sets of
502 which the host is a member. This option fails if you attempt to
503 withdraw the master node while other nodes are in the disk set
504 owner list. This option cancels all resyncs running on the node. A
505 cluster reconfiguration process that is removing a node from the
506 cluster membership list effectively withdraws the host from the
507 ownership list.
508 -y
511 Execute a subsequent take. If the take operation encounters
513 ``tagged data,'' the take operation exits with code 2. You can then
514 run the metaset command with the -q option to see an enumerated
515 list of tags.
516
519 Example 1 Defining a Disk Set
520 This example defines a disk set.
523 # metaset -s relo-red -a -h red blue
526 The name of the disk set is relo-red. The names of the first and second
531 hosts added to the set are red and blue, respectively. (The hostname is
532 found in /etc/nodename.) Adding the first host creates the disk set. A
533 disk set can be created with just one host, with the second added
534 later. The last host cannot be deleted until all of the drives within
535 the set have been deleted.
536 Example 2 Adding Drives to a Disk Set
539 This example adds drives to a disk set.
542 # metaset -s relo-red -a c2t0d0 c2t1d0 c2t2d0 c2t3d0 c2t4d0 c2t5d0
545 The name of the previously created disk set is relo-red. The names of
550 the drives are c2t0d0, c2t1d0, c2t2d0, c2t3d0, c2t4d0, and c2t5d0.
551 There is no slice identifier ("sx") at the end of the drive names.
552 Example 3 Adding Multiple Mediator Hosts
555 The following command adds three mediator hosts to the specified disk
558 set.
559 # metaset -s mydiskset -a -m myhost1,alias1 myhost2,alias2 myhost3,alias3
562 Example 4 Purging a Disk Set from the Node
566 The following command purges the disk set relo-red from the node:
569 # metaset -s relo-red -P
572 Example 5 Querying a Disk Set for Tagged Data
576 The following command queries the disk set relo-red for a list of the
579 tagged data:
580 # metaset -s relo-red -q
583 This command produces the following results:
588 The following tag(s) were found:
590 1 - vha-1000c - Fri Sep 20 17:20:08 2002
591 2 - vha-1000c - Mon Sep 23 11:01:27 2002
592 Example 6 Selecting a tag and taking a Disk set
596 The following command selects a tag and takes the disk set relo-red:
599 # metaset -s relo-red -t -u 2
602 Example 7 Defining a Multi-Owner Disk Set
606 The following command defines a multi-owner disk set:
609 # metaset -s blue -M -a -h hahost1 hahost2
612 The name of the disk set is blue. The names of the first and second
617 hosts added to the set are hahost1 and hahost2, respectively. The host‐
618 name is found in /etc/nodename. Adding the first host creates the
619 multi-owner disk set. A disk set can be created with just one host,
620 with additional hosts added later. The last host cannot be deleted
621 until all of the drives within the set have been deleted.
622
625 /etc/lvm/md.tab
626 Contains list of metadevice configurations.
628
631 The following exit values are returned:
632 0
634 Successful completion.
636 >0
639 An error occurred.
641
644 See attributes(5) for descriptions of the following attributes:
645 ┌─────────────────────────────┬─────────────────────────────┐
650 │ ATTRIBUTE TYPE │ ATTRIBUTE VALUE │
651 ├─────────────────────────────┼─────────────────────────────┤
652 │Availability │SUNWmdu │
653 ├─────────────────────────────┼─────────────────────────────┤
654 │Interface Stability │Stable │
655 └─────────────────────────────┴─────────────────────────────┘
656
658 mdmonitord(1M), metaclear(1M), metadb(1M), metadetach(1M), metahs(1M),
659 metainit(1M), metaoffline(1M), metaonline(1M), metaparam(1M), metare‐
660 cover(1M), metarename(1M), metareplace(1M), metaroot(1M), metas‐
661 sist(1M), metastat(1M), metasync(1M), metattach(1M), md.tab(4),
662 md.cf(4), mddb.cf(4), md.tab(4), attributes(5), md(7D)
663
668 Disk set administration, including the addition and deletion of hosts
669 and drives, requires all hosts in the set to be accessible from the
670 network.
671SunOS 5.11 4 Mar 2009 metaset(1M)