Bug #80858 Make setting of READ_BACKUP to 1 an inplace ALTER TABLE operation
Submitted: 25 Mar 2016 9:00 Modified: 21 Jun 2016 12:13
Reporter: Mikael Ronström Email Updates:
Status: Closed Impact on me:
Category:MySQL Cluster: Cluster (NDB) storage engine Severity:S3 (Non-critical)
Version: OS:Any
Assigned to: CPU Architecture:Any

[25 Mar 2016 9:00] Mikael Ronström
WL#9020 introduced Read Backup on new tables.
In order to make it easier to make the feature useful for new customers it
is desirable to make it possible to set Read Backup on an existing table
without having to do an ALTER TABLE using the copy algorithm.

How to repeat:
Nothing special

Suggested fix:
Add a phase for setting Read Backup to 1 and also to set Read Backup to 0 as part of ALTER TABLE commit.
In setting it to 1 it means in commit phase to go through all ongoing TC operations with writes and ensure
that those set the flag for delayed commit report for all operations in the table. In the complete phase
one will then set the read backup flag that makes it possible to read through the backup.
[21 Jun 2016 12:13] Jon Stephens
Documented fix in the NDB 7.5.3 changelog as follows:

    It is now possible to set READ_BACKUP for an existing table
    COMMENT="NDB_TABLE=READ_BACKUP=1" (Bug #80858, Bug #23001617).
    See "Setting NDB_TABLE options in table comments", for further
    information about the READ_BACKUP option.

Also updated the indicated portion of the documentation.