Hi Urbán, I observed that if I update log4j.properties and set "log4j.logger.org.apache.kafka.clients=DEBUG" I see an ERROR swallowed in a DEBUG message:
[2022-05-23 16:23:25,272] DEBUG [Producer clientId=producer-3] Exception occurred during message send: (org.apache.kafka.clients.producer.KafkaProducer:1000) org.apache.kafka.common.errors.RecordTooLargeException: The message is 3503993 bytes when serialized which is larger than 1048576, which is the value of the max.request.size configuration. This confirms what you suggested. After adding "max.request.size=5242880" inside config/connect-distributed.properties and setting the "max.message.bytes" property of the connect-configs topic the connector now works. Thanks for the help, Ryan On Sat, May 21, 2022 at 10:21 AM Urbán Dániel <[email protected]> wrote: > Hi Ryan, > > There are some limits, as the configs are stored inside an internal > topic of Connect. So the usual message size and producer request size > limitations apply. > You can reconfigure the internal topic to allow larger messages than the > default (I think it's 1MB), and the producer max request size for the > Connect worker. > > Besides that, there is one extra bottleneck - Connect followers might > forward the task configs to the leader through the REST API. The default > request size limit of the REST client can also cause an issue, not sure > if there is a way to reconfigure that. > > I think you should be able to find some kind of a related error message, > probably in the leader of the cluster, or the worker which was hosting > the connector itself. > > Daniel > > 2022. 05. 19. 17:01 keltezéssel, Ryan Slominski írta: > > Hi, > > Are there limits to the size of configuration data passed via the > > taskConfigs method of the Connector class? I'm observing a situation > > where if I use a large configuration no tasks are created, and no log > > messages appear in the connect log file. Using a smaller configuration > > works. If there are limits, can I increase them? Also, it's probably a > > good idea if Kafka were to log a warning message of some kind in this > > scenario instead of silently failing. I'm using a custom Source > Connector > > and I have documented steps to reproduce the issue using Docker compose > > here: > > > > https://github.com/JeffersonLab/epics2kafka/issues/11 > > > > Thanks for any insights! > > > > Ryan > > > > -- > Ezt az e-mailt az Avast víruskereső szoftver átvizsgálta. > https://www.avast.com/antivirus > >
