Bug #96767 for Net-Stomp: failover: re-loop through the failover-hosts ?

Fri Jun 27 08:54:56 2014 geogeurt [...] gmail.com - Ticket created

Subject:	failover: re-loop through the failover-hosts ?
Date:	Fri, 27 Jun 2014 14:54:48 +0200
To:	bug-Net-Stomp [...] rt.cpan.org
From:	Geurt Schimmel <geogeurt [...] gmail.com>

version 0.46 with failover, it goes through the list of failover-hosts, and dies at the end of the list. Would be nice if it would start at the beginning of the hostlist again after reaching the end of the list. A typical example where this would be helpful is when ActiveMQ is shut on hostA for maintenance, Net::Stomp fails over to hostB. Subsequently, ActiveMQ on hostA is started again, and ActiveMQ on hostB is shut for maintenance. Thanks, Geurt

Fri Jul 11 05:41:48 2014 DAKKAR [...] cpan.org - Correspondence added

I can't reproduce this behaviour *after the initial connection*. When you create a Net::Stomp object, it expects at least one broker to be up. After you've created it, it will keep trying all brokers when it notices a connection error while writing. Yes, this behaviour is a bit inconsistent. I may add a constructor option to keep re-trying during the initial connection, as well.

Fri Jul 11 05:41:48 2014 The RT System itself - Status changed from 'new' to 'open'

Fri Jul 11 05:41:49 2014 DAKKAR [...] cpan.org - Taken

Fri Jul 11 10:51:24 2014 DAKKAR [...] cpan.org - Correspondence added

0.47 has C<initial_reconnect_attempt> and C<reconnect_attempt>, fixing this.

Fri Jul 11 10:51:25 2014 DAKKAR [...] cpan.org - Status changed from 'open' to 'resolved'