[LinuxFailSafe] No cluster_id entry on sec. node

[Lars Marowsky-Bree]

On 2002-07-17T06:37:17,
   Daniel Berg <daniel_berg@mail.com> said:

> I've done the changes you told me to do, so now the only thing in
> /etc/hosts are the names, shortnames and ip:s of the two nodes in
> the cluster, and the localhost-entry. 

Ok.

> > You also haven't enabled / configured STONITH for the nodes in the cluster.
> > That won't work either.
> Well, now I have done that, using the ssh-module, however this is
> not the primary issue so I can deal with this some other time, 
> when I have gotten the whole cluster going.

Well, if you haven't set it up correctly, you won't get it going ;-) The reset
configuration is _important_; otherwise FailSafe can't resolve most tied up
situations cleanly.

> Wed Jul 17 13:19:22.273 <W crsd crs 21405:0 crsd_pending.c:492> CI_CRSERR_INVAL, The node specified for monitoring has its controlled port disabled. Ignoring this request.
> Wed Jul 17 13:19:24.489 <W crsd crs 21405:0 crs_config.c:667> CI_ERR_NOTFOUND, SystemController information for node tyson not found, requests will be ignored.
> Wed Jul 17 13:19:25.583 <W crsd crs 21405:0 crsd_pending.c:492> CI_CRSERR_INVAL, The node specified for monitoring has its controlled port disabled. Ignoring this request.

Please, configure STONITH for the nodes in question and enable it.

> Wed Jul 17 13:26:27.166 <W ha_cmsd cms 21597:0 cmsd_bcast.c:142> Message (from node tore:1) with a different CDB checksum
> 	local checksums  = 0x4f1d8f674c066e88:0xc47ff2b988a07c7:0x8d98537b0060c3b4
> 	remote checksums = 0x4f1d8f674c066e88:0xc47ff2b988a07c7:0xdbc4f08e0ccc25e6,

This ought not happen.

> The local machine is tore but these logs are taken from the second
> node in the cluster, ingo, if that is of importance.

FailSafe doesn't have any distinction between "local node" and "cluster node",
that is just important for defining the first node in the cluster (which must
be the one the configuration tools are run on), and then you add the other
nodes in.


Sincerely,
    Lars Marowsky-Brée <lmb@suse.de>

-- 
Immortality is an adequate definition of high availability for me.
	--- Gregory F. Pfister