I have these notes from an IBM Redpiece. I can't find the original copy, but
here are my notes.
Doesn't answer your question of how to cancel the recoveries though. I'd be
interested in that answer myself. We have a system with almost 700 ZFS
filesystems, and IPLing that after a failure can take about 15 mins to recover
all the filesystems.
Not sure why yours is taking almost an hour though.
START OF NOTES --------------------------------------------------
2.12.2 Mounting zFS file systems copied outside the sysplex
If you have a specific system located in another sysplex that is used for
preparing and servicing UNIX System Services file systems that are type zFS,
then the first mount may take at least 65 seconds. This is caused by the write
protection implementation.
As long as the aggregate is attached read-write, the identifier is active in
block zero (0). As a result, on ADRDSSU COPY or DUMP processing, it is copied
as well.
When you mount or attach the aggregate in the target production system, zFS
sees this identifier. Because it does not reflect the sysplex environment, zFS
waits 65 seconds for an update of the time stamp value within the identifier.
If the aggregate is used read-write outside of the sysplex, then this time
stamp should get updated twice a minute, as mentioned earlier.
If no update occurs, this shows that the identifier is there without a reason.
In this case, after the 65 seconds, zFS attaches and mounts the aggregate.
To avoid the 65 seconds wait time on access to the aggregate in the production
environment, unmount or detach the source aggregate in the service system
before using ADRDSSU COPY or ADRDSSU DUMP.
OR
Switch the source aggregate to read-only access, using the ISHELL modify
interface from the mount table panel or the /usr/sbin/chmount shell command.
Then copy the contents of the aggregate. This switch removes the write
protection identifier, as well.
END OF NOTES --------------------------------------------------
We also now run the following JCL as part of our z/OS cloning process:
//ZFSADM12 EXEC PGM=IOEAGSLV,REGION=0M,
// PARM=('-aggregate SYS1.SIGYROOT.OSY7A1 -recoveronly -verifyonly')
//SYSPRINT DD SYSOUT=*
//SYSUDUMP DD SYSOUT=*
Regards
Patrick Loftus
TNT Express ICS Ltd
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
