Hi Ted, Sorry for the late response I forgot to subscribe to the Mailing list…
I read that topic but it’s seems that they loose events only while upgrading their cluster. However in my case my cluster is upgraded however I’m stuck with this format version because of one of our legacy consumers. We used this guide to upgrade: https://kafka.apache.org/documentation/#upgrade_11_0_0 <https://kafka.apache.org/documentation/#upgrade_11_0_0> but we didn’t do the final step because of our legacy consumer. Here are the protocol version and message format version we are using on all our brokers (that are all running Kafka 0.11.0.0) inter.broker.protocol.version=0.11.0 log.message.format.version=0.10.2 Did I assume something incorrectly? Thanks, Karim > On Nov 2, 2017, at 1:46 PM, Karim Lamouri <ka...@lamouri.me> wrote: > > Hi, > > I wanted to know in which cases those warning arise and how to reduce them > (getting rid of them is even better): > > ~~~~ > Attempting to send response via channel for which there is no open connection > ~~~~ > > I’m asking because I linked those warnings to data loss on my production > cluster. Indeed that’s the only kind of warning I get within 20 minutes (+/-) > of data loss (0.06% on a 20 nodes cluster). > > I’m running on Kafka 0.11.0.0 (confluent platform version) and my message > format are set to 'log.message.format.version': ‘0.10.2' > > Any help or pointers is appreciated, > > Thanks, > > Karim >
