Re: selinux and ndb - Stewart Smith - com.mysql.lists.cluster

6 messages in com.mysql.lists.clusterRe: selinux and ndb

From	Sent On	Attachments
Johnny Tan	11 Dec 2007 15:28
Stewart Smith	11 Dec 2007 16:38
Johnny Tan	11 Dec 2007 17:17
Stewart Smith	11 Dec 2007 19:35
Johnny Tan	12 Dec 2007 09:20
Stewart Smith	12 Dec 2007 18:14

Subject:	Re: selinux and ndb
From:	Stewart Smith (stew...@mysql.com)
Date:	12/12/2007 06:14:29 PM
List:	com.mysql.lists.cluster

On Wed, 2007-12-12 at 12:20 -0500, Johnny Tan wrote:

Stewart Smith wrote:

On Tue, 2007-12-11 at 20:17 -0500, Johnny Tan wrote:

[root@ck:~]# /usr/libexec/ndb_mgmd --print-full-config Error : Could not alloc node id at 172.16.8.10 port 1186: No free node id found for ndb_mgmd(MGM).

specify the config file too.

Do I need to stop the cluster to do this?

no.

I'm doing it from my ndb-mgmd box:

[root@mgm:~]# /usr/libexec/ndb_mgmd \ -f /etc/mysql-cluster.cnf -P Error : Could not alloc node id at 172.16.8.10 port 1186: Connection done from wrong host ip 172.16.8.10.

Interesting... try adding --no-defaults to avoid reading any mysql_cluster section for connectstring. My guess is some connectstring is interefering.

And, just to be clear, can I specify the same port in each [tcp] section?

only one process may listen on a tcp port at a time, so that constraint needs to be observed.

but no reason why all hosts can't have a daemon listening on port X.

I still haven't seen how it's supposed to look, but from googling it seems the use of portnumber under [tcp default] is deprecated. Is that what I'd be setting here?

not in the default section, for each [tcp] section (i.e. each connection). This means the port numbers are predictable.

What are the downsides?

longer configuration file.