The ActiveMQ Journal Component allows messages to be stored in a rolling log file and then consumed from that log file. The journal aggregates and batches up concurrent writes so that the overhead of writing and waiting for the disk sync is relatively constant, regardless of how many concurrent writes are being done. Therefore, this component supports and encourages you to use multiple concurrent producers to the same journal endpoint.
Each journal endpoint uses a different log file and therefore write batching (and the associated performance boost) does not occur between multiple endpoints.
This component only supports one active consumer on the endpoint. After the message is processed by the consumer's processor, the log file is marked and only subsequent messages in the log file will get delivered to consumers.
activemq.journal:directoryName[?options]
So for example, to send to the journal located in the /tmp/data
directory you would use the following URI:
activemq.journal:/tmp/data
You can append query options to the URI in the following format,
?option=value&option=value&...
Name | Default Value | Description |
---|---|---|
syncConsume
|
false
|
If set to true , when the journal is marked after a message is
consumed, wait till the Operating System has verified the mark update is safely stored on
disk. |
syncProduce
|
true
|
If set to true , wait till the Operating System has verified the
message is safely stored on disk. |
The consumer of a Journal endpoint generates DefaultExchange objects with the In message set as follows:
journal
header: set to the endpoint URI of the journal the message came from.location
header: set to a Location which identifies where the record was stored on disk.Message body: set to ByteSequence, which contains the byte array data of the stored message.
The producer to a Journal endpoint expects an Exchange with an In message where the body can be converted to
a ByteSequence or a byte[]
.