[
https://issues.apache.org/jira/browse/ARTEMIS-1982?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16559629#comment-16559629
]
Keith Wall edited comment on ARTEMIS-1982 at 7/27/18 12:07 PM:
---------------------------------------------------------------
I've been successful in reproducing the first part of the described defect
({{MessageCount}} etc) going negative. There is a standalone reproducer for
the issue at the following location which requires only Artemis and the
reproducer. Instructions for included in the {{README.md}}. There appears to
be an accounting problem in the Broker when the Broker receives a Modified
disposition once maximum delivery count is reached.
[https://github.com/k-wall/artemis_defect_1982].
I have verified this reproducer against 2.5.0, 2.6.2 and master (today)
Relating this back to the EnMasse case with the Standard address space. Here
the Artemis Broker(s) are fronted by one or more Qpid Dispatch Routers. In a
consuming client disconnects abnormally, the Router generates a Modified
disposition for each outstanding message. It the Broker's processing of these
dispositions that is exposing the defect.
So far I have been unsuccessful in reproducing the second part to the defect
(namely the odd message flow behaviour)
was (Author: k-wall):
I've been successful in reproducing the first part of the described defect
({{MessageCount}} etc) going negative. There is a standalone reproducer for
the issue at the following location which requires only Artemis and the
reproducer. Instructions for included in the {{README.md}}. There appears to
be an accounting problem in the Broker when the Broker receives a Modified
disposition.
[https://github.com/k-wall/artemis_defect_1982].
I have verified this reproducer against 2.5.0, 2.6.2 and master (today)
Relating this back to the EnMasse case with the Standard address space. Here
the Artemis Broker(s) are fronted by one or more Qpid Dispatch Routers. In a
consuming client disconnects abnormally, the Router generates a Modified
disposition for each outstanding message. It the Broker's processing of these
dispositions that is exposing the defect.
So far I have been unsuccessful in reproducing the second part to the defect
(namely the odd message flow behaviour)
> Queue#MessageCount negative and messages stop flowing to AMQP consumer
> ----------------------------------------------------------------------
>
> Key: ARTEMIS-1982
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1982
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Affects Versions: 2.5.0, 2.6.2
> Reporter: Keith Wall
> Priority: Major
> Attachments: ARTEMIS-1982.log
>
>
> I am running performance tests against EnMasse which currently uses Artemis
> 2.5.0. Many millions of messages have traversed the Broker (all AMQP
> protocol).
>
> I have encountered a situation where messages have stopped flowing to the
> AMQP consumer. On inspection of the Artemis console, I notice that
> statistics including the MessageCount have gone negative (-99). One in this
> state, I notice if I produce one message then attach a consumer, no message
> flows. However if add two messages to the queue, the consumer gets only the
> 2nd message. The first message is seemingly lost.
> I've seen this situation once.
>
> bash-4.2$ curl --user admin:admin -k
> [https://localhost:8161/console/jolokia/read/org.apache.activemq.artemis:broker=%22broker-0%22,component=addresses,address=%22queue%22,subcomponent=queues,routing-type=%22anycast%22,queue=%22queue%22]
>
> {code:javascript}
> {
> "request": {
> "mbean":
> "org.apache.activemq.artemis:address=\"queue\",broker=\"broker-0\",component=addresses,queue=\"queue\",routing-type=\"anycast\",subcomponent=queues",
> "type": "read"
> },
> "value": {
> "User": "null",
> "MaxConsumers": -1,
> "Address": "queue",
> "Exclusive": false,
> "PurgeOnNoConsumers": false,
> "DurableDeliveringSize": 0,
> "PersistentSize": -68112,
> "ScheduledCount": 0,
> "MessagesKilled": 99,
> "Name": "queue",
> "DurableScheduledCount": 0,
> "DurableMessageCount": 0,
> "ExpiryAddress": null,
> "ID": 6,
> "FirstMessageAsJSON": "[{}]",
> "RoutingType": "ANYCAST",
> "Paused": false,
> "DurableDeliveringCount": 0,
> "FirstMessageAge": null,
> "DeadLetterAddress": "DLQ",
> "DurablePersistentSize": 0,
> "MessagesAdded": 39470041,
> "ConsumerCount": 1,
> "DeliveringCount": -99,
> "MessagesAcknowledged": 39469942,
> "MessagesExpired": 0,
> "DeliveringSize": -68112,
> "LastValue": false,
> "Temporary": false,
> "FirstMessageTimestamp": null,
> "Filter": null,
> "ScheduledSize": 0,
> "Durable": true,
> "DurableScheduledSize": 0,
> "MessageCount": -99
> },
> "timestamp": 1532007015,
> "status": 200
> }
> {code}
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
