mkrginput script

Generates stanza files for recovery group creation from IBM Storage Scale RAID topology files.

Synopsis


mkrginput  TopologyFile1 TopologyFile2 [-s | -m ] [ --match percent ]

Availability

Available on all IBM Storage Scale editions.

Description

The mkrginput script generates the stanza files needed for the creation of IBM Storage Scale RAID recovery groups by the mmcrrecoverygroup command. It takes as input the topology files generated by the mmgetpdisktopology command on the two servers that will share the recovery groups, and generates as output two stanza files for use by the mmcrrecoverygroup command.

The mkrginput script tries to verify that the two topology files are from different servers that share the same set of disk enclosures, and will exit with an error if this is not the case.

Despite this checking, it is imperative that the topology files first be examined with the topsummary command to ensure that both of them reflect the intended IBM Storage Scale RAID disk topology. If there are any disk or cabling errors, they should be corrected, and the topology files should be reacquired with mmgetpdisktopology and verified with topsummary before they are used with mkrginput to generate the stanza files.

After the topologies have been verified to be correct, they can be supplied as command arguments to mkrginput. The mkrginput command will then generate a left and right stanza file, each named with the serial number of the first disk enclosure. The left and right recovery groups can then be created by supplying the respective stanza files and servers in two separate mmcrrecoverygroup commands.

To create recovery group stanza files, follow these steps:

On the first server, server1, run:
```
mmgetpdisktopology > server1.top
```
On the second server, server2, run:
```
mmgetpdisktopology > server2.top
```
To verify the correctness of server1's topology, run:
```
topsummary server1.top
```
To verify the correctness of server2's topology, run:
```
topsummary server2.top
```
Repeat steps 1 - 4, correcting any disk subsystem problems until both topologies reflect the intended configuration.
Run:
```
   mkrginput server1.top server2.top
   GNR server disk topology: ESS GS6 HDD
   GNR recovery group layout: GS GSS/ESS RG
   Number of recovery groups: 2
   Created stanza file: G46503NL.stanza
   Created stanza file: G46503NR.stanza
   Stanza file contains '#%vdisk:' lines with log vdisk attributes.
```
The mkrginput command identifies the server topology that was detected in the server topologies and the recovery group layout that is used for the detected topology. It prescribes that two recovery groups be created, and gives the names of the stanza files that have been created for them. The stanza files will have commented out %vdisk stanzas for use after the recovery groups have been created, with suggested values for all the required log vdisks.

At this point look in the current directory for two files, one named serialnumberL.stanza and one named serialnumberR.stanza. Suppose the first enclosure in the topologies has the serial number G46503N. The mkrginput command will have created two files: G46503NL.stanza and G46503NR.stanza. Suppose the left recovery group will be named RGL and the right recovery group will be named RGR.

To create the left recovery group, RGL, run:

mmcrrecoverygroup RGL -F G46503NL.stanza --servers server1,server2

To create the right recovery group, RGR, run:

mmcrrecoverygroup RGR -F G46503NR.stanza --servers server2,server1

It is imperative that the disk topology is not changed between the capture of the topology files with mmgetpdisktopology and the creation of the two recovery groups with mmcrrecoverygroup. No server reboots or disk enclosure changes should be performed after the topologies have been verified as correct with topsummary until after recovery group creation.

After the recovery groups have been created, the commented-out %vdisk stanzas can be extracted from the recovery group input file and edited to include the recovery group names:


   grep %vdisk G46503NL.stanza
   #%vdisk: vdiskName={rg}LOGTIP rg={rg} da=NVR diskUsage=vdiskLogTip raidCode=2WayReplication blocksize=2m size=48m
   #%vdisk: vdiskName={rg}LOGTIPBACKUP rg={rg} da=SSD diskUsage=vdiskLogTipBackup raidCode=Unreplicated blocksize=2m size=48m
   #%vdisk: vdiskName={rg}LOGHOME rg={rg} da=DA1 diskUsage=vdiskLog raidCode=4WayReplication blocksize=2m size=60g 
                      longTermEventLogSize=4m shortTermEventLogSize=4m fastWriteLogPct=90

The string {rg} should be replaced with the name of the recovery group that was created using this stanza file. The comment symbol # should be removed, and the resulting vdisk stanza file may be supplied to the mmcrvdisk command to create the required log vdisks for the recovery group. The process should be repeated for the second recovery group. See the mmcrvdisk command command for more information.

Parameters

TopologyFile1: Specifies the name of the topology file from the first server.
TopologyFile2: Specifies the name of the topology file from the second server.
-s: Creates a single large declustered array. It is the default.
-m: This is a deprecated legacy option to create multiple smaller declustered arrays. It should only be used to maintain compatibility with certain GSS and ESS server topologies where there are preexisting recovery groups with multiple declustered arrays. The default is -s.
--match percent: Specifies the minimum percentage match of the number of disks and their locations in the server topologies, as reported by the topsummary command. The default is 100, since it is not recommended to create recovery groups from imperfect server topologies. Valid values are 75 - 100, inclusive.

Exit status

0: Successful completion.
nonzero: A failure occurred.

Security

You must have root authority to run the mkrginput script.

The node on which the script is issued must be able to execute remote shell commands on any other node in the cluster without the use of a password and without producing any extraneous messages. For additional details, see the following IBM Storage Scale RAID: Administration topic: Requirements for administering IBM Storage Scale RAID.

Location

/usr/lpp/mmfs/samples/vdisk