[
https://issues.apache.org/jira/browse/ARTEMIS-1203?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
razvan tanasescu updated ARTEMIS-1203:
--------------------------------------
Description:
Not entirely sure if the following behavior is as expected or is a bug.
steps
----
start 2 servers similar to cluster-static-discovery example
but with <message-load-balancing>ON_DEMAND</message-load-balancing>
producer_1 sending messages to server1 on a queue
consumer_1 start on server 1
consumer_2 start on server 2 -> messages begin being balanced
kill consumer_2 -> for a while, consumer_1 receives every one of two messages
as if messages are being still balanced then consumer_1 consecutive messages.
restart after a long time consumer_2 -> consumer_2 receives messages balanced
on server_2
----
not sure if it's a valid usecase, but in case i never start again the
consumer_2 those messages will not be consumed.
i've noticed that if i have only one server and multiple consumers, if i kill
one of the consumers, after a while, the messages that were supposed to reach
that killed consumer are eventually consumed by the other active consumers so i
was expecting the same behavior in a cluster configuration with ON_DEMAND
message balancing.
was:
Not entirely sure if the following behavior is as expected or is a bug.
steps
----
start 2 servers similar to cluster-static-discovery example
but with <message-load-balancing>ON_DEMAND</message-load-balancing>
producer_1 sending messages to server1 on a queue
consumer_1 start on server 1
consumer_2 start on server 2 -> messages begin being balanced
kill consumer_2 -> for a while, consumer_1 receives even messages as if
messages are being still balanced then consumer_1 consecutive messages.
restart after a long time consumer_2 -> consumer_2 receives messages balanced
on server_2
----
not sure if it's a valid usecase, but in case i never start again the
consumer_2 those messages will not be consumed.
i've noticed that if i have only one server and multiple consumers, if i kill
one of the consumers, after a while, the messages that were supposed to reach
that killed consumer are eventually consumed by the other active consumers so i
was expecting the same behavior in a cluster configuration with ON_DEMAND
message balancing.
> possible loss of messages in cluster
> ------------------------------------
>
> Key: ARTEMIS-1203
> URL: https://issues.apache.org/jira/browse/ARTEMIS-1203
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Reporter: razvan tanasescu
>
> Not entirely sure if the following behavior is as expected or is a bug.
> steps
> ----
> start 2 servers similar to cluster-static-discovery example
> but with <message-load-balancing>ON_DEMAND</message-load-balancing>
> producer_1 sending messages to server1 on a queue
> consumer_1 start on server 1
> consumer_2 start on server 2 -> messages begin being balanced
> kill consumer_2 -> for a while, consumer_1 receives every one of two messages
> as if messages are being still balanced then consumer_1 consecutive messages.
> restart after a long time consumer_2 -> consumer_2 receives messages balanced
> on server_2
> ----
> not sure if it's a valid usecase, but in case i never start again the
> consumer_2 those messages will not be consumed.
> i've noticed that if i have only one server and multiple consumers, if i kill
> one of the consumers, after a while, the messages that were supposed to reach
> that killed consumer are eventually consumed by the other active consumers so
> i was expecting the same behavior in a cluster configuration with ON_DEMAND
> message balancing.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
