Bug #58849 Crash ndbmtd

• Submitted: 9 Dec 2010 21:30
• Modified: 8 Sep 2016 5:56
• Reporter: Giichiro Sasaki
• Status: Not a Bug
• Category: MySQL Cluster: Cluster (NDB) storage engine
• Severity: S2 (Serious)
• Version: mysql-5.1-telco-7.1
• OS: Linux
• CPU Architecture: Any
• Tags: mysql-5.1.47 ndb-7.1.8

[9 Dec 2010 21:30] Giichiro Sasaki

Description:
Crash ndbmtd at yesterday.

Time: Thursday 9 December 2010 - 13:51:59
Status: Temporary error, restart node
Message: Internal program error (failed ndbrequire) (Internal error, programming error or missing error message, please report a bug)
Error: 2341
Error data: SimulatedBlock.cpp
Error object: SUMA (Line: 3442) 0x00000002
Program: /opt/mysql/libexec/ndbmtd
Pid: 1772 thr: 1
Version: mysql-5.1.47 ndb-7.1.8
Trace: /var/mysql-ndb/ndb_8_trace.log.3 /var/mysql-ndb/ndb_8_trace.log.3_t1 /var/mysql-ndb/ndb_8_trace.log.3_t2 /v

How to repeat:
I donot know way to repeat this.but crash 2times duaring 4days.

[9 Dec 2010 21:44] Giichiro Sasaki

os version
centos5.4 64bit (on Xen virtual machine)

compile option
CONF_COMMAND='./configure  '\''--prefix=/opt/mysql'\'' '\''--localstatedir=/var/mysql'\'' '\''--with-plugins=ndbcluster,innob
ase,partition,blackhole,'\'' '\''--with-charset=utf8'\'' '\''--with-extra-charsets=complex'\'' '\''--with-fast-mutexes'\'' '\
''--enable-assembler'\'' '\''--enable-thread-safe-client'\'' '\''--with-readline'\'' '\''--with-big-tables'\'''

[15 Dec 2010 11:16] Hartmut Holzgraefe

Please provide the log and trace files from all management and cluster nodes. The error log entry alone is not sufficient to analyze this kind of problems.

You can either collect the logs manually or use the ndb_error_reporter tool that comes with the cluster distribution (requires working ssh access between nodes)

http://dev.mysql.com/doc/refman/5.1/en/mysql-cluster-programs-ndb-error-reporter.html

[19 Dec 2010 7:20] Giichiro Sasaki

I finished uploading all management node log & all data node log & some sql node log

[20 Dec 2010 2:14] Giichiro Sasaki

Maybe I think crashed trigger is  "alter table ,drop table "

[20 Dec 2010 13:51] Jonas Oreland

1) how many mysqld do you have attached to cluster ?
2) how many of these are generating binlog (e.g started with --log-bin)
3) do you have any ndbapi applications

/Jonas

[20 Dec 2010 14:15] Giichiro Sasaki

1) how many mysqld do you have attached to cluster ?
 12nodes - 15nodes .

2) how many of these are generating binlog (e.g started with
--log-bin)
all nodes. 12-15

3) do you have any ndbapi applications
No. all of mysqld.

[23 Dec 2010 9:01] Hartmut Holzgraefe

Do you really need all mysqld nodes to write a binlog?

Having two of them do that would already be sufficient to allow for replication channel failover.

Having all of them write a binlog causes much extra load as each of them needs to subscribe to change events on all cluster tables, and for each change the changed data needs to get transfered an extra 12-15 times which binds both data node and network resources a lot ...

[23 Dec 2010 9:17] Giichiro Sasaki

>Do you really need all mysqld nodes to write a binlog?
No.I donot need bin-log of all sqlnodes :)
Because I copied my.cnf to all nodes.

>12-15 times which binds both data node and network resources a lot ...
wow Jesus!..

Do not crash claster when turn off bin-log all mysqld?

ok.I try to it.
thanks.