Bug #33406 ndb_restore -e restores epoch with wrong (big) value from old backups
Submitted: 20 Dec 2007 15:34 Modified: 7 Jan 2008 12:09
Reporter: Tomas Ulin Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Replication Severity:S3 (Non-critical)
Version:CGE 6.2/3 OS:Any
Assigned to: Tomas Ulin CPU Architecture:Any

[20 Dec 2007 15:34] Tomas Ulin
ndb_restore -e restores to big value into apply_status if restore done from old (pre micro GCP) version, e.g. mysql-5.0/1 or CGE-6.1

How to repeat:

Suggested fix:
workaround in current CGE versions is to shift value of epoch down 32 bits to get correct value
[20 Dec 2007 16:04] Bugs System
A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:


ChangeSet@1.2683, 2007-12-20 16:36:18+01:00, tomas@whalegate.ndb.mysql.com +4 -0
  Bug #33406  	ndb_restore -e restores epoch with wrong (big) value from old backups
[7 Jan 2008 12:09] Jon Stephens
Documented in the 5.1.23-ndb-6.3.8 changelog as follows:

        ndb_restore -e restored excessively large
        values to the ndb_apply_status table's
        epoch column when restoring to a MySQL
        Cluster version supporting Micro-GCPs from an older version that
        did not support these.        
        A workaround when restoring to MCCGE releases
        supporting micro-GCPs previous to MySQL 5.1.23-ndb-6.3.8 is to
        perform a 32-bit shift on the epoch column
        values to reduce them to their proper size.
[2 Feb 2008 11:53] Jon Stephens
Also documented for 5.1.23-ndb-6.2.11.
[20 Feb 2008 16:03] Bugs System
Pushed into 6.0.5-alpha