Партнёры:
Хостинг:
iiadm - command-line interface to control Sun StorageTek Availability Suite Point-in-Time Copy operations
iiadm -e {ind | shd} master_vol shadow_vol bitmap_vol
iiadm -ne ind master_vol shadow_vol bitmap_vol
iiadm [-p] [-n] {-c | -u} {s | m} volume_set
iiadm [-adDilR] volume_set
iiadm [-p] [-n] -w volume_set
iiadm [-hilLv]
iiadm -P delay units volume_set
iiadm -P volume_set
iiadm -A overflow_vol volume_set
iiadm [-OQ] overflow_vol
iiadm -E volume_set
iiadm [-IJ] volume_set bitmap
iiadm -g group_name [-aAcdDeEilLmPRuw]
iiadm [-C] cluster_tag [options]
Point-in-Time Copy software is a point-in-time snapshot feature of the Solaris operating system.
A Point-in-Time Copy snapshot is an instantly-available, time-fixed, replicated view of a momentarily quiesced volume. Once a snapshot is taken, Point-in-Time Copy software allows immediate read/write access to both the master and shadow volume data.
Point-in-Time Copy software tracks the differences between the master and shadow volumes (caused by writes) from the moment that the snapshot was established. This capability allows applications accessing the master volume's data to move forward in time independently of applications accessing the shadow volume's data, and vice-versa.
The Point-in-Time Copy software's tracking of differences between the master and shadow volumes facilitates a fast resynchronization or a full copy at a later time. The volume resynchronization can occur from either shadow to master or master to shadow.
Instantly after the point-in-time is (re-)established (either when the CLI prompt returns or the next shell script command is read), the master volume can be remounted or the applications using them can be resumed. Also, the shadow volume can be mounted and immediately accessed.
The iiadm command line utility performs only one action per command invocation. Because of this, you cannot combine multiple options, except in combination with the following overall command modifiers:
An independent (ind) snapshot causes Point-in-Time Copy software to perform a full volume copy operation from the master to the shadow. When the copy completes, the shadow volume data is identical to the master volume data at the moment that it was established. Create an independent shadow if you require two physical copies of the data. An independent shadow volume must be the same size or greater than the size of the master volume. Sun recommends that the master and shadow volumes be the same size for environments where resynchronization from shadow to master is a consideration.
A dependent (dep) snapshot causes Point-in-Time Copy software not to perform a full volume copy. The resulting shadow volume relies on the master volume for all unmodified data blocks, which are not copied until requested. Create a dependent shadow when you do not require two physical copies of the data. A dependent shadow volume can be either the same size or smaller than the master volume. A smaller shadow volume is called a Compact Dependent Shadow Volume, and is typically used when the amount of change that occurs to a Point-in-Time Copy volume set is small compared to the entire size of the master volume.
The following syntax allows you to create an exportable independent shadow volume in a Sun Cluster environment:
# iiadm -ne ind master shadow bitmap
An issue arises when using a Compact Dependent Shadow Volume in that its size is established at the time that the Point-in-Time Copy volume set is enabled. If the amount of change to the entire volume set over the duration of its usage exceeds the space allocated for the shadow volume, the shadow volume is marked as out of space. It is possible to read from the shadow volume even after it is out of space, until a portion of the data for which there was no room is requested. Once that happens, the read fails and the shadow volume is marked offline.
To address this issue, Point-in-Time Copy supports the ability to associate an overflow volume to an existing Point-in-Time Copy dependent volume set. Thus, if the size of the Compact Dependent Shadow Volume is too small, or an unscheduled amount of change occurs to the volume set, changed data can be redirected to the associated overflow volume. To facilitate efficient usage of this overflow volume, it can be associated with multiple Point-in-Time Copy volume sets on an as-needed basis.
Prior to invoking an Point-in-Time Copy enable, copy or update operation, Point-in-Time Copy assures that the shadow volume is not mounted, to prevent a file system panic from occurring. Also, it is suggested that you either unmount or suspend (quiesce) all applications using the master volume, for only the instant when the point-in-time snapshot is taken. This assures that an atomically consistent point-in-time snapshot is taken.
It is suggested that, if the master volume was suspended rather than unmounted, the new point-in-time shadow volume's integrity be validated using volume validation utilities, such as fsck(1M). The reason is that Point-in-Time Copy has made a point-in-time copy of a mounted master volume to an unmounted shadow volume. During the mounting of the shadow volume, the file system detects that it is in the mounted state. Typically this state occurs only when a system crashes, so the file system attempts to validate the integrity of the volume assuming a system failure occurred, not an Point-in-Time Copy.
The ii_bitmap variable in the /usr/kernel/drv/ii.conf configuration file determines the bitmap volume operational semantics as follows:
0
1
If a system failure occurrs while using ii_bitmap=0, the shadow volume might be inconsistent and fast resynchronization would not be possible.
If Point-in-Time Copy is used in conjunction with the Network Storage component Remote Mirror or in a Sun Cluster, set ii_bitmap=1.
The ii_debug variable in the /usr/kernel/drv/ii.conf configuration file determines the amount of information logging that is output to the system console /dev/console during Point-in-Time Copy processing.
0
1
2
The iiadm utility supports the following options.
-e{ind|dep} master_vol shadow_vol bitmap_vol
The enable shadow set processing assures that the specified volumes are accessible, that the shadow_vol is not mounted, and that the bitmap_vol is correctly sized for the type of shadow set being created. Additionally, it assures that the volumes are under control of the SV driver ( if they are not, it puts them there), initializes the bitmap volume, and, if the volume set is an independent shadow set, a full copy operation is initiated.
On a successful enable, Point-in-Time Copy stores the specified master_vol, shadow_vol and bitmap_vol names, plus the enabling type (ind or dep), into the Point-in-Time Copy configuration store. The configuration store contains all currently configured Point-in-Time Copy Volume Sets and their associated configuration attributes. (See discussion above on independent and dependent shadow volume semantics.)
master_vol is the volume from which a point-in-time snapshot is made.
shadow_vol is the volume that contains the point-in-time snapshot.
bitmap_vol is used for tracking differences between the shadow and master volumes. When Point-in-Time Copy shadow operations are suspended or resumed, the bitmap volume (maintained in kernel memory) can be stored in or retrieved from permanent storage. The storage associated with the bitmap volume should be as redundant as that of the shadow volume storage.
The shadow_vol name is the name that the Point-in-Time Copy Shadow Set is known by for all iiadm options requiring specification of a volume_set name.
-d volume_set
If Point-in-Time Copy was running in independent mode as specified in the -e ind options, above, the shadow volume data contains the same data as it did before it was disabled (assuming no writes have occurred). Users can access the master and shadow volumes, as they are now standalone point-in-time copies.
During the time that the full copy is active, an independent volume operates as though it is a dependent volume. To assure that the volume is no longer in full copy mode, issue the following command to wait for the full copy to complete:
# iiadm -w volume_set
[-p] -u s volume_set
Updates a point-in-time copy of the master volume to the shadow volume. volume_set is the Point-in-Time Copy shadow set containing the master and shadow volumes. This option provides a fast resynchronization of the shadow volume, creating an incremental copy of the master. This update copies all 32KB segments flagged as different between the master and shadow volumes. It does not copy all master volume data, only changed data. While the data is being copied, the shadow is dependent upon the master volume.
Before using this option, momentarily quiesce the workload to the volumes; stop the host application from writing to the volumes. This ensures that the point-in-time data is consistent. You can visually check the status of this copy or update operation with iiadm -i volume_set, or interactively (by means of a shell or script) with iiadm -w volume_set, before using the target volume for any other operations.
This command supports PID (Process IDentifier) locking, by using the option -p, iiadm -p -u s. Enabling this option prevents other processes from taking a new point-in-time snapshot, thus invalidating prior point-in-time data.
[-p] [-n] -u m volume_set
Before using this option, momentarily quiesce the workload to the volumes; stop the host application from writing to the volumes. This ensures that the point-in-time data is consistent. You can visually check the status of this copy or update operation with iiadm -i volume_set, or interactively (by means of a shell or script) with iiadm -w volume_set, before using the target volume for any other operations.
This command is query enabled to prevent accidentally overwriting the data on a master volume. When this command option is used in scripts, add the -n option to prevent the query from occurring.
This command supports PID (Process IDentifier) locking, by using the option -p, iiadm -p -u m. Enabling this option prevents other processes from taking a new point-in-time snapshot, thus invalidating prior point-in-time data.
[-p] -c s volume_set
Creates a point-in-time copy of the master volume to the shadow volume. volume_set is the Point-in-Time Copy volume set containing the master and shadow. This option writes all data in the point-in-time copy of the master volume to the shadow volume. While the data is being copied from master to shadow, the shadow is dependent on the master volume.
This option performs a full volume copy. Use iiadm -u s unless the integrity of the data on the independent shadow volume is in doubt. Otherwise, use this option to synchronize the master and shadow volumes; that is, make the data on each volume match.
Before using this option, momentarily quiesce the workload to the volumes; stop the host application from writing to the volumes. This ensures that the point-in-time data is consistent. You can visually check the status of this copy or update operation with iiadm -i volume_set, or interactively (by means of a shell or script) with iiadm -w volume_set, before using the target volume for any other operations.
This command supports PID (Process IDentifier) locking, by using the -p option, iiadm -p -c s. Enabling this option prevents other processes from taking a new point-in-time snapshot, thus invalidating prior point-in-time data.
-c m volume_set
Creates a point-in-time copy of the shadow volume to the master volume. volume_set is the Point-in-Time Copy volume set containing the master and shadow volumes. This option writes all data in the point-in-time copy of the shadow volume to the master volume. While the data is being copied from the shadow to the master, the master is dependent upon the shadow volume.
This option performs a full volume copy. Use iiadm -u m unless the integrity of the data on the independent master is in doubt. Otherwise, use this option to synchronize the master and shadow volumes; that is, make the data on each volume match.
Before using this option, momentarily quiesce the workload to the volumes; stop the host application from writing to the volumes. This ensures that the point-in-time data is consistent. You can visually check the status of this copy or update operation with iiadm -i volume_set, or interactively (by means of a shell or script) with iiadm -w volume_set, before using the target volume for any other operations.
This command is query-enabled to prevent accidentally overwriting the data on a master volume. When this command option is used in scripts, add the -n option to prevent the query from occurring.
This command supports PID (Process IDentifier) locking, by using the -p option, iiadm -p -c m. Enabling this option prevents other processes from taking a new point-in-time snapshot, thus invalidating prior point-in-time data.
-a volume_set
[-p] [-n] -w volume_set
This option waits until the current Point-in-Time Copy operation is complete, thus preventing a subsequent iiadm command (from a shell or script) from executing. Use this command option when you need to be sure the copy or update operation has completed.
This command supports PID (Process IDentifier) unlocking. If a prior copy or update, using a command iiadm -p {-c|-u} {m|s}, was invoked with the -p option, upon completion of the wait processing, if the current PID was the PID that locked the point-in-time data, this option unlocks the data.
-i volume_set
-l
-O overflow_vol
During initialization of the overflow_vol, the initiator of this option, must answer the following question: "Initialize this overflow volume? yes/no" A response of either "yes/no" is required before proceeding.
This option supports the -n option, so that the requested action is performed without prompting. This option is useful for inclusion in a script. The -n option must be specified first. For example, "iiadm -nO vol" is valid; "iiadm -On vol" is not.
Make sure you want to initialize the data on the specified overflow_vol, especially when using the -n option.
-A overflow_vol volume_set
If the overflow_vol has not been initialized, this option initializes the overflow_vol (see -O option), then attaches the overflow_vol to the volume_set.
If overflow_vol was previously initialized, this option attaches the overflow_vol to the volume_set.
This option supports the -n option, so that the requested action is performed without prompting. This option is useful for inclusion in a script. The -n option must be specified first. For example, "iiadm -nA vol" is valid; "iiadm -An vol" is not.
Make sure you want to initialize the data on the specified overflow_vol, especially when using the -n option.
-D volume_set
abort(-a)
disable(-d)
update(-u)
-L
-Q overflow_vol
-E volume_set
-I volume_set bitmap_vol
After the exported/imported independent shadow volume is no longer needed by the other node, you must enter a disable command so that the bitmap_vol and its associated shadow_vol are consistent, prior to performing a join operation. For example,
# iiadm -d volume_set
-J volume_set bitmap_vol
-g group_name -m volume_set [volume_set2 ...]
Only the commands COPY (-c) and UPDATE (-u) are performed atomically across all Point-in-Time Copy sets within the group. All other grouped, iiadm commands are performed sequentially on each member of the group.
The syntax of an iiadm group command is as follows:
iiadm -g group_name [options]
The options are as follows:
-a
-A
-c {s | m}
-D
-d
-E
-i
-l
-L
-n
-P
-R
-u {s | m}
-w
-g -" " -m volume_set [volume_set2 ...]
-C cluster_tag
In a Sun Cluster where the volume manager is Sun Cluster-aware, iiadm automatically obtains the correct Disk Group information, therefore this option is typically not required unless the volumes are part of an encompassing Resource Group.
In a Sun Cluster where the volumes are accessible on the local node only, the special cluster_tag of local is used to indicate volumes that are not part of a Sun Cluster Resource Group or Disk Group.
If "-L" is given as a the cluster_tag argument, then iiadm lists all cluster tags associated with Point-in-Time Copy.
This option is invalid when used on a Solaris system on which the Sun Cluster package has not been installed or configured.
-h
-v
Contact Sun Enterprise Services for assistance in using the remaining commands in this section.
-P delay unit volume_set
-R volume
0
>0
See attributes(5) for descriptions of the following attributes:
| |||||||||
dscfg(1M), svadm(1M), ds.log(4), rdc.cf(4), attributes(5), ii(7D), sv(7D)
|
Закладки на сайте Проследить за страницей |
Created 1996-2024 by Maxim Chirkov Добавить, Поддержать, Вебмастеру |