Do you eventually run OOM? On Sunday, February 28, 2016, Asaf Mesika <asaf.mes...@gmail.com> wrote:
> Hi, > > I'm seeing slow off-heap memory leak in production. > I've managed to recreated the scenario in a testing environment - I have > one ZooKeeper node and one Kafka Broker (running on same machine). I have > one java process which runs a thread which constantly writes to Kafka using > 16 KafkaProducer. > I see that the resident memory of my java process is slowly decreasing > (roughly 250mb every 15 min). When running GC it cleans the heap to 730mb > roughly but resident remains high: 2.3gb and increasing steadily. So from > that I presume it's a leak of off-heap memory. > > I'm little puzzled here since I haven't seen any usage of > ByteBuffer.allocateDirect() in the new client code, but I though I'll check > with the audience in case you are aware of any issue that might cause it? > I'm chasing this leak for several days, and managed to track it down to the > code writing to Kafka, so I'm a little desperate :) any help will do. > > Thanks! > > Asaf > -- *Christian Posta* twitter: @christianposta http://www.christianposta.com/blog http://fabric8.io
