Saving a System with BRMS When It Has a Geographic Mirrored Environment

To obtain a valid system save of a geographically mirrored environment and avoid a resync at the end of the save, the following procedure is suggested for use with BRMS. The intent behind this procedure is to quiesce the IASP and ensure that no files are currently being written during the save. This cannot be done via a function 21 save because the IASP cannot be touched without TCP and clustering active.

Notes:

1.	Start with an Active geographically mirrored environment (IASP varied on, clustering active, and geographically mirrored Active).
2.	Steps 3 through 6 are optional - if you are only saving the IASP (not SYSBAS), you do not need to restrict the system. You just need to quiesce the IASP and ensure that no users can access it after varying it back on.

To obtain a *FULL save of the production copy IASP without requiring resync, do the following:

1.	Vary off the IASP (from the production copy). This will quiesce the mirror copy and ensure it is fully insync.
2.	End the CRG. This will ensure that no accidental failovers occur.
3.	End clustering on the nodes you intend to save.
4.	Restrict the system you are planning to save.
5.	Save SYSBAS by using the SAVSYS command.
6.	Restart the system. Do not restart QINTER or any other subsystems that would allow users or jobs to access the IASP.
7.	Restart TCP/IP.
8.	Start clustering on all nodes. Start the CRG.
9.	Vary on the IASP. A resync will not be required since we were able quiesce and go insync in Step 1.
10.	Perform a *FULL save of the IASP (from the production copy). Since no users are using the IASP at this point, a full save should be successful.
11.	Restart the rest of your subsystems and allow users to access the IASP.

Note that the PRODUCTION copy is being saved rather than the mirror copy. The user cannot access the mirror copy of the IASP. All access to the IASP must be through the production copy. This means that the production copy is not available for use during the save process.

To do this for BRMS, refer to the following example of a control group:

10 *EXIT VRYCFG CFGOBJ(IASPNAME) CFGTYPE(*DEV) STATUS(*OFF))
20 *EXIT (Add CL to end the CRG here)
30 *EXIT (Add the CL to end clustering on the nodes you want to save)
40 *EXIT ENDSBS *ALL *IMMED (see Note)
50 *SAVSYS (My guess is there are more entries here as they probably want User libraries, and so on, also)
60 *EXIT Not sure what subsystems, but they could be re-started here)
70 *EXIT STRTCPSVR
80 *EXIT (Add the CL command to start clustering)
90 *EXIT Start the CRG
100 *EXIT VRYCFG CFGOBJ(IASPNAME) CFGTYPE(*DEV) STATUS(*ON))
110 *ALLUSR IASPNAME Type of SWA if needed.
120 *LINK IASPNAME
130 *EXIT STRSBS (controlling subsystem) (see Note)

Note: BRMS can be allowed to end and restart the subsystems using Option 9 on the control group. If this option is used, these exits are not required.