MySQL Bugs: #9826: Server crash on schema change ("drop table", "alter table") with NDB

Bug #9826	Server crash on schema change ("drop table", "alter table") with NDB
Submitted:	11 Apr 2005 21:03	Modified:	4 Jun 2005 10:22
Reporter:	Joerg Bruehe
Status:	Duplicate
Category:	Server: Cluster	Severity:	S1 (Critical)
Version:	5.0.6	OS:	HP/UX (HP-UX 32 bit, ppc linux)
Assigned to:	Martin Skold	Target Version:

Description:
Build of 5.0.4, based on ChangeSet
  1.1889 05/04/08 16:49:14 mskold@mysql.com +4 -0
  Merge from fix of Bug #9675  Auto-increment not working with INSERT..SELECT and NDB
storage

Test 'ndb_alter_table' fails on both 'hpux11' and 'hp3750' (each: 32 bit only, both
default and PS protocol run) with this symptom:
=== quote start ===
ndb_alter_table                [ fail ]

Errors are (from
/var/tmp/mysqldev/hpux11/test/mysql-max-5.0.4-beta-hp-hpux11.11-hppa2.0w/mysql-test/var/log/mysqltest-time)
:
/var/tmp/mysqldev/hpux11/test/mysql-max-5.0.4-beta-hp-hpux11.11-hppa2.0w/bin/mysqltest:
At line 126: query 'drop table t1' failed: 2013: Lost connection to MySQL server during
query
(the last lines may be the most important ones)

Ending Tests
=== quote end ===

Subsequent test failures seem to be related or consequences: 
ndb_autodiscover, ndb_autodiscover2, ndb_basic, ndb_bitfield, ndb_blob, ndb_cache2,
ndb_charset, ndb_condition_pushdown, ndb_index_ordered, ndb_index_unique, ndb_limit,
ndb_lock, ndb_multi, ndb_read_multi_range, ndb_restore, ps, ps_7ndb, query_cache, rename,
select, show_check, system_mysql_db, system_mysql_db_fix, view.

How to repeat:
Build + test.

Not a high priority platform for cluster

With 5.0.6, the crash happens even earlier:

ndb_alter_table                [ fail ]

Errors are (from
/home/mysqldev/hp3750/test/mysql-cluster-5.0.6-beta-hpux11.00-hppa2.0w/mysql-test/var/log/mysqltest-time)
:
/home/mysqldev/hp3750/test/mysql-cluster-5.0.6-beta-hpux11.00-hppa2.0w/bin/mysqltest: At
line 21: query 'ALTER TABLE t1 ADD COLUMN c int not null' failed: 2013: Lost connection
to MySQL server during query
(the last lines may be the most important ones)

In addition to "hpux11" and "hp3750" (both 32 bit, both "cluster" + "max"), the same test
failure also happened on "pegasos2".

I am wondering if this is a similar issue to bug 10948 (in the ndb replication tree)

I did have (what I believed to be) a replication only problem, which I then believed I
fixed. I now don't think I have (as I can still get the tests to fail). But maybe it's
just this bug now?

I have reproduced on ppc linux.

Reproduced with latest 5.0-ndb tree.

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/internals/25559

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/internals/25604

Please do not submit the same bug more than once. An existing
bug report already describes this very problem. Even if you feel
that your issue is somewhat different, the resolution is likely
to be the same. Because of this, we hope you add your comments
to the original bug instead.

Thank you for your interest in MySQL.

Additional info:

Complete information contained in BUG#10948.