Re: problem start mgm node - Jon Stephens - com.mysql.lists.cluster

29 messages in com.mysql.lists.clusterRe: problem start mgm node

From	Sent On	Attachments
Mikael Ronström	05 May 2007 01:34
Robin Bowes	05 May 2007 08:55
Sebastien LELIEVRE	15 May 2007 05:42
Johan Andersson	16 May 2007 02:57
Sebastien LELIEVRE	16 May 2007 03:04
Johan Andersson	16 May 2007 03:15
Sebastien LELIEVRE	16 May 2007 03:33
Sebastien LELIEVRE	16 May 2007 08:46
Mikael Ronström	16 May 2007 13:06
Tangirala, Srikalyan	16 May 2007 13:18
Matthew Montgomery	16 May 2007 21:29
Tangirala, Srikalyan	17 May 2007 06:15
Matthew Montgomery	17 May 2007 07:45
Tangirala, Srikalyan	17 May 2007 08:26
Matthew Montgomery	17 May 2007 08:52
Tangirala, Srikalyan	17 May 2007 11:06
Sebastien LELIEVRE	18 May 2007 01:59
Sebastien LELIEVRE	18 May 2007 02:29
Rosario Pingaro	20 May 2007 20:45
Jon Stephens	20 May 2007 23:30
Rosario Pingaro	21 May 2007 00:59
Jon Stephens	21 May 2007 02:53
Sebastien LELIEVRE	21 May 2007 03:09
Jonathan Miller	25 May 2007 06:00
Tangirala, Srikalyan	29 May 2007 08:50
Geert Vanderkelen	29 May 2007 09:55
Tangirala, Srikalyan	29 May 2007 11:13
Tangirala, Srikalyan	29 May 2007 13:16
Stewart Smith	31 May 2007 03:47

Subject:	Re: problem start mgm node
From:	Jon Stephens (jo...@mysql.com)
Date:	05/21/2007 02:53:51 AM
List:	com.mysql.lists.cluster

Rosario Pingaro wrote:

You are right without connect string it started fine. But when i started the cluster more then a year ago I had to use connect string to have the two mam node working at the same time.

AFAIK, you must reference all management servers when starting data nodes and SQL nodes, but not when starting management nodes.

cheers

jon.

Regards Rosario

Rosario Pingaro wrote:

I had to restart a mgm node and now when i start the mgm daemon i got the following error:

[root@cs1 ~]# ndb_mgmd -f /var/lib/mysql-cluster/config.ini --connect-string=10.0.0.10,10.0.0.11 Could not alloc node id at 10.0.0.11 port 1186: Connection done from wrong host ip 10.0.0.10.

The configration didn't change.

I run cluster 5.0.18 and it was fine since last year.

I don't think a connectstring is even required for management hosts - try starting it without one and see what happens.

Any help? Regards