Hi Alex,<div><br></div><div>Thank you for the detailed feedback.</div><div>Meanwhile I have found and fixed some issues on the software making use of RabbitMQ.<br>I made a tool which has a work pool of multiple processes running (Python/multiprocessing/Pika).<br>Under certain circumstances a worker crashes while the connection to the broker stays alive, which left the message in a permanent unacknowledged state as long the work pool manager stayed alive.</div><div>This has been fixed. I'm afraid this situation has clouded my judgement on finding the reason for this behavior.<br><br></div><div>Here are the relevant portions of version info. (All brokers run the same version)</div><div><br></div><div>... snip ...</div><div><div>{erlando,"Syntax extensions for Erlang","2.6.1"},</div><div> {rabbitmq_federation,"RabbitMQ Federation","2.6.1"},</div><div> {rabbitmq_shovel,"Data Shovel for RabbitMQ","2.6.1"},</div><div> {rabbitmq_shovel_management,"Shovel Status","2.6.1"},</div><div> {rabbitmq_management,"RabbitMQ Management Console","2.6.1"},</div><div> {webmachine,"webmachine","1.7.0-rmq2.6.1-hg0c4b60a"},</div><div> {rabbitmq_management_agent,"RabbitMQ Management Agent","2.6.1"},</div><div> {amqp_client,"RabbitMQ AMQP Client","2.6.1"},</div><div> {rabbit,"RabbitMQ","2.6.1"},</div><div> {os_mon,"CPO CXC 138 46","2.2.6"},</div><div> {sasl,"SASL CXC 138 11","2.1.9.4"},</div><div> {rabbitmq_mochiweb,"RabbitMQ Mochiweb Embedding","2.6.1"},</div><div> {mochiweb,"MochiMedia Web Server","1.3-rmq2.6.1-git9a53dbd"},</div><div> {inets,"INETS CXC 138 49","5.6"},</div><div> {mnesia,"MNESIA CXC 138 12","4.4.19"},</div><div> {stdlib,"ERTS CXC 138 10","1.17.4"},</div><div> {kernel,"ERTS CXC 138 10","2.14.4"}]},</div></div><div>... snip ...</div><div><br></div><div>I will upgrade my environment to the latest stable version 2.7.0 as this is the latest stable release.</div><div><br></div><blockquote class="gmail_quote" style="margin-top: 0px; margin-right: 0px; margin-bottom: 0px; margin-left: 0.8ex; border-left-width: 1px; border-left-color: rgb(204, 204, 204); border-left-style: solid; padding-left: 1ex; ">since a broker restart<br><span class="Apple-style-span" style="text-align: left; "> will mark the unacknowledged messages as ready again, and deliver<br></span><span class="Apple-style-span" style="text-align: left; "> them to a consumer.</span></blockquote><div><br></div><div>In my case I was unable to restart the broker, ... it just hung indefinitely forcing me to kill the RabbitMQ process.</div><div><br></div><div>What I will do is the following:</div><div><ul><li><span class="Apple-style-span" style="line-height: normal;">Upgrade all participating RabbitMQ servers to the latest stable version (2.7.0)</span></li><li><span class="Apple-style-span" style="line-height: normal;">Try to reproduce the problem.</span></li><ul><li><span class="Apple-style-span" style="line-height: normal;">If the problem occurs I provide you detailed and complete information broker configuration (I rather do this offline then)</span></li><li><span class="Apple-style-span" style="line-height: normal;">If the problem can't be reproduced, I'll provide feedback on that to this post.</span></li></ul></ul></div><div>Thanks again Alex for the time taken to read through an analyse the provided information, much appreciated.<br></div><div><br></div><div><br></div><div>Cheers,</div><div><br></div><div>Jelle</div><div><br></div><div><br></div><div><br></div><div><br></div>