MySQL Forums
Forum List  »  NDB clusters

Unstable Cluster alpha release
Posted by: Ryan Perc
Date: December 06, 2007 11:42AM

Hello All,

I am testing Version:6.0.2 alpha and am running into multiple problems. Some may be due to my current configuration, or possible due to the nature of the alpha version.

I am running a 4 node cluster with 2 mgm nodes, here is the config (same on both mgm nodes)

[NDBD DEFAULT]
DataMemory = 768M
IndexMemory = 32M
RedoBuffer = 512M
NoOfReplicas = 2
MaxNoOfOrderedIndexes = 600
MaxNoOfAttributes = 3000

[MYSQLD DEFAULT]

[NDB_MGMD DEFAULT]

[TCP DEFAULT]
SendBufferMemory = 2M


# Section for the cluster management node
[NDB_MGMD]
HostName=10.0.0.1
Id=1
ArbitrationRank=1

[NDB_MGMD]
HostName=10.0.0.3
Id=2
ArbitrationRank=2

# Section for the storage nodes
[NDBD]
HostName=10.0.0.1
DataDir= /var/lib/mysql-cluster
Id=11

[NDBD]
HostName=10.0.0.2
DataDir= /var/lib/mysql-cluster
Id=12

[NDBD]
HostName=10.0.0.3
DataDir=/var/lib/mysql-cluster
Id=13

[NDBD]
HostName=10.0.0.4
DataDir=/var/lib/mysql-cluster
Id=14

# one [MYSQLD] per storage node
[MYSQLD]
HostName=10.0.0.1
Id=21

[MYSQLD]
HostName=10.0.0.2
Id=22

[MYSQLD]
HostName=10.0.0.3
Id=23

[MYSQLD]
HostName=10.0.0.4
Id=24

Here is the my.cnf
[mysqld]
ndbcluster
ndb-connectstring= 10.0.0.1,10.0.0.3

[MYSQL_CLUSTER]
ndb-connectstring=10.0.0.1,10.0.0.3

Now what I am trying to do is capture all server events in the database. All email and apache logs are redirected via pipes and server error logs via syslog-ng.

The main issue I see is when I take any of the 4 nodes down (similating a failure), the cluster gets "confused" and will not let me preform any inserts/selects on the databases until all api nodes are up. If I restart mysql and wait for the "failing" api node to join everything works correctly.

I have at times not been able to get the api nodes up at all, short of stopping and restarting all related processes.

Typically the errors I get in the logs are related to inserts/selects to a database that is not "up" due to a api node being down.

Can you let me know what I can do to try to isolate the issues? Maybe there is something I should look at in the configurations?

Thanks,
Ryan



Edited 1 time(s). Last edit at 12/06/2007 12:23PM by Ryan Perc.

Options: ReplyQuote


Subject
Views
Written By
Posted
Unstable Cluster alpha release
2050
December 06, 2007 11:42AM
1271
December 10, 2007 12:25AM
1269
December 10, 2007 11:01AM
1216
December 10, 2007 11:05AM
1282
December 10, 2007 04:09PM


Sorry, you can't reply to this topic. It has been closed.

Content reproduced on this site is the property of the respective copyright holders. It is not reviewed in advance by Oracle and does not necessarily represent the opinion of Oracle or any other party.