Out of operation records in local data manager far below MaxNoOfConcurrentOperations


we run a MySQL-Cluster, currently version 7.4.15, where we get "Out of
operation records in local data manager" at approximately 55000 rows
updated or deleted in one transaction,
even though MaxNoOfConcurrentOperations=300000.

Any ideas, why 300000 operation records are not enough for 55000 rows?

Hendrik Woltersdorf

FinTech Group AG

Q: OPTIMIZE TABLE - internal error 122 during operation


We have an NDB cluster and we experience 'mysterious' database-full errors.

Our setup is the following:

- server_mgm: Management Node
- server_1: Data Node 1
- server_2: Data Node 2

(Host names are explanatory.)


# tuning
MaxNoOfConcurrentTransactions = 8192
MaxNoOfConcurrentOperations = 65536
MaxNoOfConcurrentScans = 500
# end of tuning



回复:Re: Sharding key of secondary index with respect to sharding key of the table and shard by range?

Thanks Mauritz.

Here is what I understand you ( all for NDB engine ):
1. No requirement for partitioning columns to be included in secondary indexes, regardless of unique index or non-unique index.
2. For any secondary indexes, there are usually two parts, one is as hash index, another is a tree index. The hash index is sharded independently, so any updates involving secondary index will usually require distributed transaction, so a performance penalty. The tree part of a secondary index is co-located with main table, and is not co-located with the hash part.

Sharding key of secondary index with respect to sharding key of the table and shard by range?

I have two questions, related.
1. As far as I know, mysql has support for secondary index, which is good, but how does mysql shard a secondary index. The doc says secondary index is implemented as index table, then is the index table sharded independently of the main table?

can not insert data in mysql cluster


I met a trouble when inserting data to MySQL cluster.

The error is "The table 'admin_log' is full".
(admin_log is a table name)

Increase innodb_file_per_table seems no help.

My question is "Is there any related parameter can be increased to solve this problem ?"

Thank you very much for help.


mysql cluster and VPN

hi !

We are wondering if a mysql cluster witch nodes will be replicated in
tow "groups nodes" distributed between tow sites could work fine
(performance and latency) in "VPN connexion" (in order to get the
cluster in a isolated network as recommanded for sécurity)

thanks for all advises


回复:Re: Re:Re:_Re:Re:_auto-inc_column_as_secondary_field_to_compose_a_primary_key_in_mysql_cluster

Thank you Mauritz.

On 2016-11-20 07:53, JohnXhark wrote:


Hi Mauritz,

Thanks for the extra info.

top (n) is equavalent to mysql syntax "limit n".

One more question, if I use id as pk (and thus as sharding key automatically by mysql cluster), then, for a query like "select * from tbl where id > id1 order by id limit n",
Will Mysql Cluster limit the query request to a single data node or a small sub-set of nodes?


Thank you Jesper, it clarifies a lot.


----- 原始邮件 -----
发件人:Jesper Wisborg Krogh < ... at wisborg dot dk>
收件人: ... at lists dot
日期:2016年11月17日 16点14分

Hi John,
On 17/11/2016 16:40, JohnXhark wrote:

Re:Re: auto-inc column as secondary field to compose a primary key in mysql cluster

Hi Mauritz,
Thanks for your answering.
I found this from the doc:
" InnoDB table with an AUTO_INCREMENT column requires at least one key where the auto-increment column is the only or leftmost column."

So my index definition will not work for innoDB, and it will not work as expected for MyISAM as you just said. it seems this is a mess. But, specifically, will it work as expected for NDB engine?

The motivation for my design is this:
we have a use case, heavily query of the pattern "select top (n) * from tbl where cityid = some_city and id > id1 and ...

Re:Re: auto-inc column as secondary field to compose a primary key in mysql cluster

Hi Mauritz,

Thanks for your answering.

I found this from the doc:
" InnoDB table with an AUTO_INCREMENT column requires at least one key where the auto-increment column is the only or leftmost column."

So my index definition will not work for innoDB, and it will not work as expected for MyISAM as you just said. it seems this is a mess. But, specifically, will it work as expected for NDB engine?

The motivation for my design is this:
we have a use case, heavily query of the pattern "select top (n) * from tbl where cityid = some_city and id > id1 and ...

auto-inc column as secondary field to compose a primary key in mysql cluster

I would like to define a primary key like this: [shardingkey, id(auto-inc)], is this possible? Here the id is a secondary field in primary key, but not the primary key by itself.

However, it's not clear if this is legitimate, the document at says:
AUTO_INCREMENT columns. Like other MySQL storage engines, the NDB storage engine can handle a maximum of one AUTO_INCREMENT column per table. However, in the case of a Cluster table with no explicit primary key, an AUTO_INCREMENT column is automatically defined and used as a “hidden” primary key.

NDB for multiple independent SQL nodes

Hello everyone,
I have a little question about the captabilities of the NDB System.

My scenario is the following:

A single NDB cluster composed by same data nodes and management nodes,
plus 2 sql nodes.

Now what I would do is logically partition the cluster, so the SQL node
1 cannot see the databases and the users of the SQL node 2, even if they
are still using the same NDB cluster as storage.

Can this scenario be realized or it's a totally unsupported feature?

Thank you for your comments about it
Best regards

What is the equivalent in mysql cluster for global table in mysql fabric?

What is the equivalent in mysql cluster for global table in mysql fabric, i.e. for table that are too small to need any partition/sharding?
My guest is just define a table of innodb engine and copy it to every sql node. But is this correct?
Can tables of different engines be joined in a mysql node?

Creating Custom Distributed Storage Engine

Hi All,

I saw the documentation here:

<a href="" title=""></a>

But I wasn't sure if this documentation was valid for the case where I want
to create a distributed, shared, storage backend for MySQL, where many
MySQL servers are all accessing the same storage backend.

Could someone please point me in the right direction?

Lost correct information after restart sql node

Hello all,

I had a cluster scenario (1 mgmt node, 1 sql node, 2 data nodes) working
Today, after a firewall configuration between nodes, I tried to restart

SQL node could not restart, it's logging:

2016-04-14 10:23:52 17805 [ERROR] /usr/sbin/mysqld: Incorrect information
in file: './mysql/db.frm'
ERROR: 1033 Incorrect information in file: './mysql/db.frm'
2016-04-14 10:23:52 17805 [ERROR] Aborting

But, when I installed, I ran script that correct privileges and change
tables to ndbcluster engine.

Does anyone know what to do?

I tried to run mysql_install_db, but problem cont

NDB API multithreading issue


I have tried to use NDB API directly in order to obtain better performance,
and results are tremendous. Thumbs up!

However, I have faced an interesting issue that happens only on Windows,
while on Linux everything works just fine. It is a simple C++ application
where I perform initialization in one thread, and read data in another. In
other words, if I do everything in a single thread - it works. The error I
get is "write access violation _my_thread_var(...) returned nullptr".

Reorganize Partition On Partial Cluster

I have changed MAX_ROWS on partial (while a data node disconnected) cluster
then reorganized partitions.

bit data type was not full supported by MySQL cluster ?

I create a tale in MySQL cluster 7.4.4,and one column is bit data type.

when I use sub-query in a SQL statement with the ‘where’ condition on the bit data type,the query result is not the excepted.

My table script is:

CREATE TABLE `rt_urls` (

`id` int(11) NOT NULL AUTO_INCREMENT COMMENT 'primary key',

`serverid` varchar(50) NOT NULL COMMENT 'server ID',

`s_check_urls_id` int(11) NOT NULL COMMENT ‘setting置ID',

`url` varchar(255) NOT NULL COMMENT 'access Url',

`enabled` bit(1) NOT NULL DEFAULT b'0' ,

`msg` varchar(2000)

Error Occured on Creating Logfile

Dear MySQL team,

May I reach for assistance regarding with our setup of MySQL Cluster.

About MySQL Cluster storage limit

According to
<a href="" title=""></a>
, the storage limit of MySQL Cluster is 3TB. I also see that 48 * 64GB is
exactly 3TB.

I know from
<a href="" title=""></a>
l that the maximum number of data nodes in a cluster is limited to 48.

However, it's not clear whether 3TB is the upper limit of the amount of RAM
that NDB/MySQL Cluster can manager, or whether the storage limit of a MySQL
cluster can be increased by adding more RAM to the data nodes. i.e.

Mysql Cluster Data Node Hangs on Phase 5 While Starting

I have an mysql cluster which have 4 api nodes, 2 management nodes and 4
data nodes.

Mysql Cluster Node failure caused abort of transaction Errors


I have an mysql cluster which have 4 api nodes, 2 management nodes and 4
data nodes. Today, I was having problems while trying to connect database
and all queries was hanging at "Opening tables" state.

performance issues with clusterj


I was wondering about clusterj behavior while increasing number of
concurrent transactions.

In my test application, i have a number of worker threads each instantiated
with a Session object, and each run many transactions in sequence. Each
transaction reads 5 rows from the database (primary key operations) using
read-committed lock type. Each worker thread reads different set of rows.

I tried first to vary number of worker threads within my test process, as
well as adding multiple test processes in parallel.

ALTER TABLE foo Engine=NDB performance


I did some test experiments with MySQL Cluster and noticed some unexpected
performance penalty while writing to a remote ndb.

1. sql and ndb node are on the same host

"alter table engine=ndb" needed about 4 minutes for about one million
rows. After enabling another data node the data has been transferred to
the other node with 40Mbits/s.

2. sql and ndb node are on different hosts, network latency is about 0.3ms

"alter table engine=ndb" needed more than 30 minutes.

Create MySql cluster on Hyper-v

Hello all, When creating MySQL cluster using Auto-installer on Hyper-v , and trying to add the local machine as a host, it fail to get the resource information of the machine giving me [error : Number_of_Processors] , also, doesn't give the right MySQL install directory , and MySQL data directory. Is this related to creating cluster using hyper-V and not on physical machine?
Thanks and best regards

Eng. Sara Abd El-Razek Borghol
System & Security engineer
Scientific computing center
Helwan universityMob: +20 114 300 32 35
Tel: 2555 89 45
2555 89 17
Fax: 2555 94 91

Re: ndb_restore failed due foreign key violation


Mysql cluster version is 7.4.5 (community server).

On Mon, May 18, 2015 at 9:22 AM, Kallol Chakraborty <

ndb_restore failed due foreign key violation


I am trying to restore mysql cluster backup on slave cluster. Master and
slave cluster have identical configuration. (2 api nodes, 2 management
nodes and 4 data nodes, same system resources)

When i run ndb_restore command it gives me following error:

$ /usr/local/mysql/bin/ndb_restore -n 5 -b 15 -r ./5 -m


Opening file './6/BACKUP-15.6.log'

File size 1210464 bytes

execute failed: 255: Foreign key constraint violated: No parent row

How can i restore cluster backup safely?

Re: Attempting to install mysql cluster on RHEL7.1


From EPEL, python-paramiko-1.15.1-1.el7.noarch


I also had to remove the mariadb-lib in rhel7.1 as that is compatible with the above.


Hello Steven,

How did you install paramiko ? by yum installer or downloaded the zipped version.

Attempting to install mysql cluster on RHEL7.1


When I run the auto-installer gui I get a no module named paramiko?

I have python-paramiko installed from EPEL but no joy.

Can I do or do I need to do a sym link? if so to where? ie where is the auto installer looking?



My First Cluster - Getting error on Nodegroup configuration

Hello list. I am new to this list and new to MySQL Cluster.

I have been having an issue with setting up a small sample cluster on both CentOS 6.6 and 7.1. Both OS’s give me the same error.

Essentially, I am copying the steps outlined on the MySQL Cluster installation page.

7.2 RPM repackaging thoughts ...

Starting with MySQL Cluster 7.2 there don't seem to be separate
'storage' and 'management' RPMs anymore; 'ndbd', 'ndbmtd' and
'ndb_mgmd' binaries are now part of the 'server' RPM along with

As the upgrade order for nodes during a rolling upgrade is
"management nodes, then data nodes, and only in the last step
mysqld/API nodes" I see the following problem with this new
combined RPMs:

* it is not uncommon to run ndb_mgmd or ndbd and mysqld on the
same machines in parallel,

* to do a rolling upgrade i need to replace ndb_mgmd binaries
first, as these are not in a separate package

Compatibility information in the manual not matching source code?

AFAIR the upgrade/downgrade compatibility information on manual
sections like:

<a href="" title=""></a>

was generated from the information found in the source file


but when looking at the 7.2 documentation I see a downgrade
compatibility break between 7.2.13 and 7.2.14 while the
source says:

struct NdbUpGradeCompatible ndbCompatibleTable_full[] = {

and doesn't have any 7.2.x specific entry in ndbCompatibleT

Is there Mysql cluster MGM API for Java?

I know there is a Mysql cluster MGM API for C, so I want to know if there is Mysql cluster MGM API for Java?
The MGM api exampls are:
ndb_init / ndb_mgm_set_connectstring / ndb_mgm_connect / ndb_mgm_get_status /ndb_mgm_get_status2



Adding management node to running ndb cluster?


I’ve been looking trough the documentation and searching the web for a howto on adding more management nodes to a NDB cluster.

So far I’ve found absolutely nothing that describes a procedure for this.

Do any of you have any links that describe this?

Best regards,
Asle Ness

sql_thread went into offline in galera cluster


we are using galera cluster in two data centers like C!,C2 .

each data center(c1/c2 ) contains 3 nodes.

C1:(galera cluster group)


C2:(galera cluster group)


we configured normal mysql replication in b/w two galera cluster nodes
like A----> F ,E---->B

Recently we are facing a issues with sql thread in slave.

sql thread went into offline automatically we didn't find any error
messages,after wards we are starting it manually and it catching fine.

*why sql thread went into offline with out any error message in Galera .*

Replication with gtid


I have 3 ndb-clusters over different regions. I have atm circular
mysql-replication with EPOCH. I want to switch to gtid, but i could not
find any documentation about ndbcluster + gtid-replication.
Does anyone tried it yet?

may I have a document for my 'mysql clusting 'excise

there I have read some document from the Internet , but some problem is very difficult for me to solve , thank for your attention!
system OS : redhat 5.4
mysql 5.1.7


ANNOUCEMENT mariadb galera on windows (compiled using msvc)

Hi this is my first post here:
A Mariadb galera 5.5.53a port compiled entirely using msvc 1600  has been posted
to <a href="" title=""></a>
It compiles right now on both x86 and amd64 hosts and is  being tested on Windows8.1
(single node right now)

Feel free to test it
Suggestions and feedback is welcome

Replication, Epoch not getting updated


I've a working circular replication between 3 NDB-clusters. for the
channel-failover i want to use epoch, since last time it worked like a
This time on one cluster the EPOCH is not getting updated.
eq. when i insert entries in each region, 2 Regions will have a new
max(epoch) but one doesnt change.. then the failover results in an error
since the MASTER response to this epoch with nothing

I dont understand it, the replication is working fine. Only this one
cluster has a problem with its EPOCH

Thanks for Help,

Clarification Required

Dear All,

On executing ndb_mgm> show, we used to see < id=1 @
(mysql-5.5.34 ndb-7.2.14, Nodegroup: 0, Master)>, but below log shows '*'
instead of Master.

Please help me to understand the difference.

Centos Mysql Cluster Installation Package

Dear All,

I want to install "MySQL Cluster" on CentOS 6.3, but the download options on
the MySQL website do not have any official - rpms for CentOS.

Does RHEL-6 rpms work for CentOS or I have to use "Generic Linux" version.

Please suggest the appropriate package to be used for CentOS 6.3.

Best Regards,

Syed Shahanavaz

Senior Software Engineer

Futures Business Development


dual-channel replication


I setup 3 ndbclusters with circular replication between the clusters,
everything working fine so far.

circular replication between clusters


I got so far that when i do "mysql>start slave;" i get:
Slave_IO_Running: Yes
Slave_SQL_Running: Yes

also POS and BINLOGFILE are correct,

but i got this output only on the node where i executed "start slave"; so i
tried to start one after another(3 sides)
but in the end at the 1st where it was working initially, it was failing in
the end when i did "start slave" at the 3rd

I am new to this so many thanks for advice!

data-nodes mgmt-node mysqld-nodes

I have just 4 APP server in one region, so my idea for now to install a
mgmt-node with pacemaker for HA + 2-3 Data-Nodes + each APP-server will
also act as mysqld-node

Since the mysqld is using the data-nodes there should not be much more load
on the APP-server?


Data retrieval causes query times to spike


Running MySQL Cluster 7.2 with 2 API (MYSQLD)/MGM nodes, 4 Data nodes (64

We are running into an issue when a job that runs every few minutes is
grabbing changed data (determined by triggers) and dumping it out to disk.
During this process, we often hit a 5 second timeout that we have
configured for MySQL Cluster, causing a MYSQL HAS GONE AWAY error.

Raising the timeout is not really an option we're looking for currently. We
have 60 API connections split between our two MYSQLD nodes.

Error writing BLOB data to cluster

Hello everyone,

New to this list, as I'm setting up aMySQL Cluster and having problems
inserting existing data into it.
The cluster seems to work correctly, and creating"simple" tables with
text data is ok. Both of my nodes have the data, and I can access it
through two SQL nodes.

But I'm trying to import data from an already running MySQL instance to
my MySQL Cluster, and it fails.
My table stores BLOB values.

CRUD in MySql general log


One of the MySql Cluster manuals says the following:
Typically a JPA implementation would access the database (for example,
MySQL Cluster) using JDBC.

RAM usage of MySQL cluster SQL nodes


I have a cluster at version 7.2.8 where the two SQL nodes use _very_
different amounts of RAM.

One 1.4GB, the other 12GB as shown by top:

top - 13:04:48 up 255 days, 2:02, 1 user, load average: 0.38, 0.40, 0.38
Tasks: 203 total, 1 running, 202 sleeping, 0 stopped, 0 zombie
Cpu(s): 1.7%us, 0.5%sy, 0.0%ni, 97.6%id, 0.0%wa, 0.0%hi, 0.2%si,
Mem: 28825436k total, 28488168k used, 337268k free, 363136k buffers
Swap: 4192956k total, 176k used, 4192780k free, 13334488k cached

19297 mysql

ClusterJDatastoreException (possible obscure Windows error?)


Hopeing someone can help me gain some insight into the following problem.

I get the following exception on my Windows setup, but not on my Linux

INFO: Created cluster connection 'localhost:1186' with node id 0.
<openjpa-2.2.2-r422266:1468616 nonfatal general error>
org.apache.openjpa.persistence.PersistenceException: Error getting
connection to cluster with properties
com.mysql.clusterj.connect.retries=4, com.mysql.clusterj.connect.delay=5,


Dear all,

Is it possible to imitate the behavior of SQL's LIMIT in NDB C++ API?

Kind regards,

Mazen Alhajri

Senior Software Engineer

Futures Business Development

P.O. Box 53870, Jeddah 21593

+966-2-6615666 Ext: 106


<a href="mailto: ... at fbd dot net"> ... at fbd dot net</a>

<a href="" title=""></a>

CONFIDENTIALITY NOTICE - This email is intended for the sole use of the
intended recipient(s) and may contain confidential or privileged
information. No one is authorized to copy or re-use this email or any
information contained in it.

Disk based tables

Hi (again):

Currently, our cluster is hosting several databases. One of the
databases is growing and growing and growing ...

Problem restarting node


I have an stupid problem with my MySQL cluster. It's version

I have two nodes and was working perfect. But, I need to change the
DataMemory value.

SO, i've changed it on the config.ini file, killed the ndb_mgmd daemon
and restarted with the --reload option.

Then, I've entered to the ndb_mgm tool and tried to restart the second
node so it gets the new config.

Problem restoring mysqldump

Hello to all,

I'm having trouble returning a backup made by mysqldump into a two-node
The dumpfile size is 177M and execute the command for its restoration
received the error "ERROR 1297 (HY000) at line 251: Got temporary error
1234 'REDO log files overloaded (Increase disk hardware)' from NDBCLUSTER."

The cluster contains two nodes each with two cores and Xeon 2.0Ghz 2Gb RAM
on a VMware Storage SATA disks in Raid 5.

Already changed NoOfFragmentLogFiles, RedoBuffer, LongMessageBuffer and
error persists.

Cluster configuration was made with the tool Severalnines.

connect to data node


New to MySql Cluster - how can I connect to the data node to see what's on
there? I want to visually see the partitions data. Or can one query MySql
somehow so that it shows you from which data node each row originated?

Many thanks,
Serdyn du Toit

Possible to auto-restart ndbd after network disconnection ?


Assume the following scenario:
-2 ndb nodes, and one management node each on separate machines.
-I unplug node 2: It detects that it cannot reach node 1 and tries to contact the management node.

Re:is there anyway to auto_extend a datafile?

here is my config.ini






is there anyway to auto_extend a datafile?


i am a newbie of mysql cluster.
i set up a cluter from "mysql-cluster-gpl-7.2.10-debian6.0-x86_64.deb"

here is the problem when i tested the 'disk data table'.
i dump a database named 'news' from my standard sqlserver,it only contained one table named 'news_cr_news' and its size is 2.4G
-rw-rw---- 1 mysql mysql 2.4G 2011-02-26 news_cr_news.MYD.
so i create a tablespace that contained two datafiles,one's size is 1G,the other 2G, and i thought that was enough.
when i loaded data into the cluster, i got an error as follow:
ERROR 1114 (HY000): The table

ERROR 1528 (HY000): Failed to create UNDOFILE


I have mysqlcluster setup and when I try to run the following command:

create logfile group log_group add undofile 'log_group1.log'
initial_size=1073741824 undo_buffer_size=31457280 engine=ndbcluster;

It throws up this error:

*ERROR 1528 (HY000): Failed to create UNDOFILE *
and the warnings are:

-> ;
| Level | Code | Message
| Warning | 1296 | Got error 1509 'Fil

Syndicate content