Bug #45154 Incorrect handling of create/drop with different #partitions in DD-tables

• Submitted: 28 May 2009 12:09
• Modified: 29 May 2009 7:22
• Reporter: Jonas Oreland
• Status: Closed
• Category: MySQL Cluster: Disk Data
• Severity: S3 (Non-critical)
• Version: *
• OS: Any
• Assigned to: Jonas Oreland
• CPU Architecture: Any

[28 May 2009 12:09] Jonas Oreland

Description:
create table t1 max_rows = ABIGNUMBER storage disk;
insert 100 rows;
force lcp;
drop table t1;
create table t1 storage disk;
insert 100 rows;
system restart;

Then the system restart will examine the extra partitions that was
created due to the max_rows = ABIGNUMBER, but performs checks
on them prior to verifying that the actually exists.

This can cause signal 11.

How to repeat:
new test prg
sometimes seen in testSystemRestart -n basic

Suggested fix:
move check of fragment exists to correct place

[28 May 2009 12:43] 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:

  http://lists.mysql.com/commits/75141

2939 Jonas Oreland	2009-05-28
      ndb - bug#45154 - check frag existance before trying to access it in Dbtup::restart_alloc_extent

[28 May 2009 12:54] 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:

  http://lists.mysql.com/commits/75143

2940 Jonas Oreland	2009-05-28
      ndb - bug#45154 - add testcase to autotest

[28 May 2009 13:21] Bugs System

Pushed into 5.1.34-ndb-6.3.26 (revid:[e-mail]-20090528125413-shv6vq33lrs2db1f) (version source revid:[e-mail]-20090528125413-shv6vq33lrs2db1f) (merge vers: 5.1.34-ndb-6.3.26) (pib:6)

[28 May 2009 13:22] Bugs System

Pushed into 5.1.34-ndb-7.0.7 (revid:[e-mail]-20090528131222-31rqlof4kza88gie) (version source revid:[e-mail]-20090528131222-31rqlof4kza88gie) (merge vers: 5.1.34-ndb-7.0.7) (pib:6)

[28 May 2009 13:23] Bugs System

Pushed into 5.1.34-ndb-6.2.18 (revid:[e-mail]-20090528125333-ke9f3bro36p1wn74) (version source revid:[e-mail]-20090528125333-ke9f3bro36p1wn74) (merge vers: 5.1.34-ndb-6.2.18) (pib:6)

[28 May 2009 16:10] Jon Stephens

Documented bugfix in the NDB-6.2.18, 6.3.26, and 7.0.7 changelogs as follows:

        Having NDBCLUSTER tables created with very large values for
        MAX_ROWS could cause mysqld to crash when performing a system
        restart. This was because the server attempted to examine all
        partitions whether or not they actually existed.

[28 May 2009 16:20] Jonas Oreland

it's ndbd that might crash,
and only if the table with lots of partitions was dropped
and a new table with fewer partitions was created that got same
table id

[29 May 2009 7:22] Jon Stephens

Updated changelog entry per comment from Jonas.