mysql cluster管理点中start命令的用法

coocal

看了mysql的邮件列表（
http://lists.mysql.com/cluster/3431
），终于知道了mysql cluster管理点中start命令的用法，一个困惑我很久的问题终于有答案了。下面是邮件的原文：


原邮件：
Hi,
After I stop one ndb (node_id=2) using ndb_mgm command : 2 stop, I failed to bring it back
through ndb_mgm.
I first tried  '2 start' ,   get error msg:
'Start failed.
*    22: Error
*        No contact with the process (dead ?).
I then tried '2 restart -n' , get error msg:
Restart failed.
*  2003: Restart failed
*        5000-No contact with the process (dead ?).
The processes representing node_id=2 indeed are gone due to the '2 stop'.
mgm's show :
ndb_mgm> show
Cluster Configuration
---------------------
[ndbd(NDB)]     2 node(s)
id=2 (not connected, accepting connect from 10.89.51.242)
id=3    @10.89.51.243  (Version: 5.0.19, Nodegroup: 0, Master)
[ndb_mgmd(MGM)] 1 node(s)
id=1    @10.87.20.107  (Version: 5.0.19)
[mysqld(API)]   2 node(s)
id=4    @10.89.51.242  (Version: 5.0.19)
id=5    @10.89.51.243  (Version: 5.0.19)
Can anyone point to me what is wrong ?
Thanks,
Frank
回复一：
Hi Frank,
On Apr 5, 2006, at 17:12:48, Frank Xie wrote:
..
>
> The processes representing node_id=2 indeed are gone due to the '2  
> stop'.
..
If you 'stop' a data node, the angle process is killed too. So, the  
ndb_mgm can't can't connect anymore to start or restart it and you  
need to manually start ndbd again.
Use '2 restart' instead of '2 stop' and '2 start'.
Best regards,
Geert
回复二：
Thanks, Geert.
Based on what you said, it seems to me that the only way to start ndb is manual way, since
both 'start' and 'restart' need  the ndb process  to be there.  Sounds like  'egg first vs
chicken first'
If so, under which occasion 'start' command is used ?  
Best Regards,
Frank
Geert Vanderkelen stripped> wrote: Hi Frank,
On Apr 5, 2006, at 17:12:48, Frank Xie wrote:
..
>
> The processes representing node_id=2 indeed are gone due to the '2  
> stop'.
..
If you 'stop' a data node, the angle process is killed too. So, the  
ndb_mgm can't can't connect anymore to start or restart it and you  
need to manually start ndbd again.
Use '2 restart' instead of '2 stop' and '2 start'.
Best regards,
Geert
回复三：
Hi Frank,
On Apr 5, 2006, at 17:56:11, Frank Xie wrote:
> Thanks, Geert.
>
> Based on what you said, it seems to me that the only way to start  
> ndb is manual way, since both 'start' and 'restart' need  the ndb  
> process  to be there.  Sounds like  'egg first vs chicken first'
On a side node: egg or chicken first... Egg ofcourse, it's evolution :)
>
> If so, under which occasion 'start' command is used ?
ndb_mgm> 9 restart -n
-n means don't start the node right away, just stop it. This will not  
kill the angle process of the data node.
ndb_mgm> 9 start
This will then start the node. Could be useful.
Best regards,
Geert



本文来自ChinaUnix博客，如果查看原文请点：http://blog.chinaunix.net/u/30557/showart_292530.html