Learn the basic steps for downgrading from
Adaptive Server 15.0.3 Cluster Edition to a 15.0.1 Cluster Edition, and 15.0.1 Cluster Edition
ESDs. Additional actions are required if you used 15.0.3 Cluster Edition features.
Follow the steps in the
Additional Steps If You Used 15.0.3 Cluster Edition Features Before Downgrading topic.
Adaptive Server 15.0.3 Cluster Edition provides a procedure called sp_downgrade that
is used to help with the downgrade process. This procedure requires “sybase_ts_ role” to
be on. The “sa_role” or “sso_role” permissions
are required to run this procedure. For detailed information
on this system procedure and its syntax see “The New
System Procedures” section in the ‘System Changes” chapter
of the Adaptive Server New Feature Guide for 15.0.2.
-
If you have used Pluggable Component Interface (PCI)
functionality, and are downgrading to a version of 1502 ESD,
drop the sybpcidb and restart the old server
binary.
-
If you are downgrading to version 15.0.3, rerun installpcidb and
restart the old server binary.
-
Start the 15.0.3 Cluster Edition server in single user mode,
with only one instance running, using dataserver
-m option
to guarantee that no other user can access Adaptive Server while
you are performing the downgrade. See Utility
Guide for more details on starting the server in single
user mode.
-
Make sure Adaptive Server 15.0.3 Cluster Edition is prepared for
downgrade by executing:
sp_downgrade 'prepare', <'version'>
, @override = 1
The value of “version” may be written as “15.0.1.5” or “15015”. However, the
version entered should be the Adaptive Server version you want to downgrade
to. sp_downgrade 'prepare' validates
the readiness of Adaptive Server 15.0.3 Cluster Edition for downgrade. This means
that sp_downgrade 'prepare' may print messages
regarding manual changes required before the downgrade can be completed.
Errors reported must be corrected and implications of warnings reported
must be understood before you can proceed with further downgrade
steps. Refer to the Additional Steps If You Used 15.0.3 Cluster Edition Features Before Downgrading topic for more information on correcting
the issues reported by sp_downgrade 'prepare'.
-
Repeat step 3 until sp_downgrade
'prepare' reports no errors. Your system is now ready
to downgrade.
-
Begin the downgrade process by executing:
sp_downgrade 'downgrade', @toversion = '15015', @override = 1
The version number should be 15.0.1.5 or 15015. Once the above command
completes successfully there can be no more activity on the 15.0.3
CE server. You must immediately shutdown the 15.0.3 CE server. sp_downgrade
'downgrade' may report that the password for sa and other login
accounts has been reset and that the account is locked. In this
case, the new password shown by sp_downgrade is
used as the password for sa after the downgrade. Any other affected
login accounts should be unlocked after the downgrade is complete.
-
Run - checkpoint and shutdown the old server.
-
Copy the RUN_SERVER file
to the 15.0.1 Cluster Edition and 15.0.1 Cluster Edition ESDs release area to which you are downgrading.
Modify the RUN_SERVER file to use
the dataserver binary from the 15.0.1 Cluster Edition and 15.0.1 Cluster Edition ESDs release. Remove the -m option
when you modify the RUN_SERVER file
or ensure the server is
not
running
in single user mode.
-
Extract the qurum device using:
qrmutil --quorum-dev=<dev> --extract-config=quorum.out
-
Restart the old server using the modified RUN_SERVER file.
Note: If a downgraded server is booted using the 15.0.2 configuration
file, the new options will cause the message ‘Unknown parameter ‘.
The unknown option will be reported only the first time the server
is booted.
-
Rebuild the new quorum device with the quorum device
backup you created in step 8 by adding this to the dataserver parameters:
--buildquorum=force --cluster-input=quorum.out
-
Depending upon whether you are starting from a fresh
installation or from a previously upgraded 15.0.2 installation, run
previously saved restore_msgs.ebf or instmsgs.ebf from
15.0.1 if none were saved, before running installmaster for
the Cluster Edition.
-
Restart all instances of the cluster.