Frame body is too large

[vaibhav-kaup]

Describe the bug

spring boot java service having com.rabbitmq:amqp-client:jar:5.21.0 deployed on AWS fargate, trying to connenct to rabbitmq:3.12-management deployed via
DockerFile ->

FROM rabbitmq:3.12-management # tried rabbitmq:4.1.0-management also

RUN mkdir -p /etc/rabbitmq/conf.d
RUN chown -R rabbitmq:rabbitmq /etc/rabbitmq/conf.d
COPY logging.conf /etc/rabbitmq/conf.d
EXPOSE 5672
EXPOSE 15672

logging.conf file content :
log.console = true
log.console.level = debug
log.file.level = debug
log.connection.level = debug

spring boot service has properties
spring.rabbitmq.host=https://rabbitmq.test.com
spring.rabbitmq.port=5672
spring.rabbitmq.username=guest
spring.rabbitmq.password=guest
rabbitmq.queue.name=somefunction
rabbitmq.exchange.name=someorganization
rabbitmq.routing.key=abc.xy.#

spring boot dependency version in pom
<java.version>17</java.version>
<spring-cloud.version>2023.0.2</spring-cloud.version>
<spring-boot.version>3.3.1</spring-boot.version>

org.springframework.boot spring-boot-starter-amqp

springBootJavaSvcLogs.txt

Reproduction steps

1.deploy rabbit mq task on AWS Fargate, check its management site is working on port 15672
2.deploy spring boot java service having dependency on om.rabbitmq:amqp-client:jar:5.21.0 on AWS Fargate
3.as java service boots up it tried to connect to rabbit MQ
the same configuration works on our local unix boxes, although we use docker container for rabbit mq and java svc runs as a java cmd
...

Expected behavior

from local unix box working setup logs, it Connects and creates a new connection to rabbit MQ from the java svc

2025-06-04 07:18:40,598 INFO [main] CachingConnectionFactory: Attempting to connect to: [localhost:5672]
2025-06-04 07:18:40,785 INFO [main] CachingConnectionFactory: Created new connection: rabbitConnectionFactory#7f8f120c:0/SimpleConnection@554bb231 [delegate=amqp://[email protected]:5672/, localPort=44308]
2025-06-04 07:18:40,822 DEBUG [main] RetryTemplate: Retry: count=0
2025-06-04 07:18:40,822 DEBUG [main] RabbitAdmin: Initializing declarations
2025-06-04 07:18:40,829 INFO [main] RabbitAdmin: Auto-declaring a non-durable, auto-delete, or exclusive Queue (returns) durable:false, auto-delete:false, exclusive:false. It will be redeclared if the broker stops and is restarted while the connection factory is alive, but all messages will be lost.
2025-06-04 07:18:40,846 DEBUG [main] CachingConnectionFactory: Creating cached Rabbit Channel from AMQChannel(amqp://[email protected]:5672/,1)
2025-06-04 07:18:40,859 DEBUG [main] RabbitTemplate: Executing callback RabbitAdmin$$Lambda$2247/0x00007fc8b8d85ae0 on RabbitMQ Channel: Cached Rabbit Channel: AMQChannel(amqp://[email protected]:5672/,1), conn: Proxy@18e5b50 Shared Rabbit Connection: SimpleConnection@554bb231 [delegate=amqp://[email protected]:5672/, localPort=44308]
2025-06-04 07:18:40,859 DEBUG [main] RabbitAdmin: declaring Exchange 'someExchange'
2025-06-04 07:18:40,877 DEBUG [main] RabbitAdmin: declaring Queue 'someQueuue'
2025-06-04 07:18:40,879 DEBUG [main] RabbitAdmin: Binding destination [returns (QUEUE)] to exchange [stockloan] with routing key [sss.sl.#]
2025-06-04 07:18:40,883 DEBUG [main] RabbitAdmin: Declarations finished

Additional context

No response

[vaibhav-kaup]

relevant error in logs
java.lang.IllegalStateException: Frame body is too large (1345270062), maximum configured size is 67108864. See ConnectionFactory#setMaxInboundMessageBodySize if you need to increase the limit.

a message header of 1.3 gb is too big, we dun't know how this is generated.
we tried increasing the aws service task deployment size from 1 CPU, 3 GB Ram to 2 cpu 16GB too

also we tried to set spring.rabbitmq.max-inbound-message-body-size=1500000000, but it went into out of Memory bounds

[michaelklishin]

@vaibhav-kaup you publish concurrently on a shared channel, which results in an incorrect frame interleaving on the wire.

[basic.publish 1] [content headers 1] [content body 1]*
[basic.publish 2] [content headers 2] [content body 2]*

becomes

[basic.publish 1] [basic.publish 2] [content headers 1]  [content headers 2] [content body 1]*  [content body 2]*

or something like that, that the server does not expect and does not consider a valid frame sequence.

This is explicitly prohibited for every client except for the most recent .NET client major.

[vaibhav-kaup]

@michaelklishin thanks for replying.
This issue is on the service startup, so my code is not even started publishing to the rabbit MQ server yet.
the setup works on my local dev unix machines, with no extra configuration.
Is there a property/parameter to disable the publishing concurrently on a shared channel OR ?
please give some direction.

[michaelklishin]

@vaibhav-kaup the error message contains a hint:

Frame body is too large (1345270062), maximum configured size is 67108864.

See ConnectionFactory#setMaxInboundMessageBodySize if you need to increase the limit.

1345270062 bytes is 1.3 GiB. I somewhat doubt that it is a genuine message of that size
but if it is, then the error message tells you exactly what you need to do. Usually, however, messages are several orders of magnitude smaller.

Our team does not recommend publishing such large messages. Put them in a blob store and publish their identifiers.

[acogoluegnes]

@vaibhav-kaup Maybe a wire capture would help.

A user reported a similar error recently, it ended up being unrelated to the size of messages, see this comment.