[
https://issues.apache.org/jira/browse/ARTEMIS-5574?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Justin Bertram updated ARTEMIS-5574:
------------------------------------
Description:
In previous versions of the broker when the first bytes of a large message are
sent to the broker we would add a "pending" large message record to the journal
so that we could keep track of the message and potentially clean up any
stranded large message files (e.g. if the broker crashed before the client
finished sending the message). After upgrading these records may still exist in
the journal and they are never removed which means that *every* time the broker
is restarted thereafter it will log something like the following:
{noformat}
AMQ221005: Deleting pending large message as it was not completed:
Pair[a=<journal-record-id>, b=<large-message-id>]{noformat}
was:
When the first bytes of a large message are sent to the broker we add a
"pending" large message record to the journal so that we can keep track of the
message and potentially clean up any stranded large message files if the broker
crashes before the client finishes sending the message. However, there are 2
problems:
* The process of deleting the file actually creates it if it doesn't exist.
This creates unnecessary disk I/O.
* The "pending" large message record is never removed from the journal which
means that *every* time the broker is restarted thereafter it will log
something like the following:
{noformat}
AMQ221005: Deleting pending large message as it was not completed:
Pair[a=<journal-record-id>, b=<large-message-id>]{noformat}
> Fix handling of pending large messages
> --------------------------------------
>
> Key: ARTEMIS-5574
> URL: https://issues.apache.org/jira/browse/ARTEMIS-5574
> Project: ActiveMQ Artemis
> Issue Type: Bug
> Reporter: Justin Bertram
> Assignee: Justin Bertram
> Priority: Major
>
> In previous versions of the broker when the first bytes of a large message
> are sent to the broker we would add a "pending" large message record to the
> journal so that we could keep track of the message and potentially clean up
> any stranded large message files (e.g. if the broker crashed before the
> client finished sending the message). After upgrading these records may still
> exist in the journal and they are never removed which means that *every* time
> the broker is restarted thereafter it will log something like the following:
> {noformat}
> AMQ221005: Deleting pending large message as it was not completed:
> Pair[a=<journal-record-id>, b=<large-message-id>]{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
For further information, visit: https://activemq.apache.org/contact
