[Gluster-users] Connected but not connected?

[Whit Blauvelt]

Hi Pranith,

I got past the problem by stopping glusterd on both systems, deleting the
files from /etc/glusterd/peers, restarting glusterd, then doing the peer
probe again. Everything went normally that time. So it must have been
something about the values in the files in /etc/glusterd/peers that was in
error. 

Now I see that "State: Accepted peer request (Connected)" was not really the
response to want from "gluster peer status." With the working setup, the
response is "State: Peer in Cluster (Connected)," which makes more sense.

The not-working state before had followed it taking several peer probes to
get that step to succeed, so I'd guess the flawed files in
/etc/glusterd/peers were related to probing more than once. After deleting
those files, probing worked the first time. 

Whit

On Thu, May 19, 2011 at 04:40:35AM -0500, Pranith Kumar. Karampuri wrote:
> Hi Whit,
>     Could you please zip the logs and send them across. They are generally located in /usr/local/var/log/glusterfs/
> I also need the following outputs:
> On white:
> gluster fsm log black
> On black:
> gluster fsm log white
> 
> We need to fix the error message saying that the "peer is not in cluster".
> 
> Pranith
> ----- Original Message -----
> From: "Whit Blauvelt" <whit.gluster at transpect.com>
> To: gluster-users at gluster.org
> Sent: Thursday, May 19, 2011 6:49:53 AM
> Subject: [Gluster-users] Connected but not connected?
> 
> Hi,
> 
> This is an attempt to recreate a small replicated storage that was
> previously working between these two systems using 3.2.0, which has been
> removed from the systems (for prudential reasons) and 3.1.3 installed.
> Everything from 3.2.0 was removed _except_ that the two blocks I'm trying to
> bind still have their replicated data from 3.2.0. 
> 
> With two hosts, "white" and "black":
> 
>   root at white# gluster peer status
>   Number of Peers: 1
> 
>   Hostname: black
>   Uuid: 6f7238a2-df00-464e-a296-4bb38ddc8b1c
>   State: Accepted peer request (Connected)
>   root at white# gluster volume create bind replica 2 transport tcp white:/mnt/gluster/bind black:/mnt/gluster/bind
>   Host black not connected
> 
> Okay, black is "Connected" but "not connected." Let's try from black:
...