qpid-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alan Conway <acon...@redhat.com>
Subject Re: How to detect the initial synchronization of brokers in cluster
Date Fri, 22 Jul 2011 15:46:22 GMT
On 07/22/2011 09:38 AM, Jakub Scholz wrote:
> Hi Alan,
> When the cluster is already running, it shows the ACTIVE state
> correctly. But it doesn't work for me (MRG 1.3 ... next week I will be
> hopefully able to try it with MRG 2.0 / Qpid 0.10). And I don't think
> it can work like that even in newer versions. The state of the cluster
> is maybe "JOINING" internally, but since the qpid-cluster tool is
> trying to get the data from the broker over QMF, he doesn't get them.
> It seems to connect fine, but as with the other QMF tools, it doesn't
> get any responses since the broker isn't processing anything coming
> from the connection (due to the JOINING state).
> It seems to me as a kind of "chicken and egg" problem ... you cannot
> get the information that the cluster is in JOINING state because it
> may be in JOINING state. It is a kind of a side-effect of using AMQP
> to manage the AMQP broker :-o.

Agreed with all of the above.

> At the end, the only way would be probably to fix the timeout issue in
> the QMF (eventually write some script with Qpid API - that seems to
> wait nicely). Then the qpid-cluster tool can be used in a following
> way:
> a) Connection error =>  Not running
> b) Connects and hangs/waits =>  JOINING state
> c) Returns info that the state is ACTIVE =>  it is ACTIVE

How do you distinguish cases a) and b) ? The connection could hang/wait either 
because the remote host is down or because it is joining. I don't see how to 
distinguish the two. I suppose you could use some other means to ping the remote 
host and be sure it is reachable.

Apache Qpid - AMQP Messaging Implementation
Project:      http://qpid.apache.org
Use/Interact: mailto:users-subscribe@qpid.apache.org

View raw message