MySQL Forums
Forum List  »  NDB clusters

problem with SQL node on Solaris OS
Posted by: Ilya
Date: March 24, 2006 09:12AM

I have a trouble with MySQL Cluster being up.

I installed and configured MySQL Cluster for 3 hosts:
host2 and host3 as ndbd nodes and
host1 as mgm and sql nodes.

OS config:
host1: SunOS 5.8 Generic_117000-01 sun4u sparc SUNW,Sun-Fire-880;
host2, host3: SunOS 5.10 Generic_118822-27 sun4u sparc SUNW,Sun-Fire-V440

MySQL Cluster is up successfully. It was be possible to see by mgm client
"ndb_mgm" and command "show". All 4 nodes (2 ndbd, 1 mgm and sql) were
connected.

MySQL Cluster was being shutdown due to MySQL Documentation: on server host1 (by root) stop management node and sql server
i. ndb_mgm -e shutdown
ii. mysqladmin shutdown

After that I tried to restart MySQL Cluster folling next steps:
I. On server host1 (by root) start management node:
ndb_mgmd -f /var/lib/mysql-cluster/config.ini
II. On server host2, host3 (by root) start ndbd node:
ndbd
III. On server host1 (by root) start mysql server :
mysqld --user=mysql &

And I got trouble with starting sql:
060324 17:26:26 InnoDB: Started; log sequence number 0 0
060324 17:26:26 [ERROR] Fatal error: Can't open and lock privilege tables: Table 'mysql.host' doesn't exist

I use "/etc/init.d/mysql start" and sql server starts, but when I verify by management client 3 nodes (mgm and 2 ndbd) are connected
successfully but sql node can't work properly: ndb_mgm shows that sql node
is not connected.
ndb_mgm> show
Connected to Management Server at: localhost:1186
Cluster Configuration
---------------------
[ndbd(NDB)] 2 node(s)
id=2 @10.50.3.18 (Version: 5.0.19, Nodegroup: 0, Master)
id=3 @10.50.3.36 (Version: 5.0.19, Nodegroup: 0)

[ndb_mgmd(MGM)] 1 node(s)
id=1 @10.50.3.2 (Version: 5.0.19)

[mysqld(API)] 1 node(s)
id=4 (not connected, accepting connect from 10.50.3.2)


I tried to revive MySQL Cluster but mysql server doesn't work. These attempts included: changing configurations, clearing data source, reinstalling mysql package on each server, removing mysql package from servers at all and install it again and
shutdown servers with restarting.

Currently we have working mysql server but ndb_mgm shows that mysql node is not connected to cluster.

I tried to change data (create tables and insert data) by mysql client but
directories with data changed on server with mysql only.

Can anybody explain me:
1. Maybe some lock files exist in OS Solaris as on Linux?
2. Why ndb nodes don't connected to started sql server?
3. I need working MySQL Cluster, how to revive it?

Options: ReplyQuote


Subject
Views
Written By
Posted
problem with SQL node on Solaris OS
2254
March 24, 2006 09:12AM
1535
March 29, 2006 02:12AM


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.