Hi, RequestTimeout is used for 2 cases : 1) Timing out the batches sitting in the accumulator. 2) Requests that are already sent over the wire and you have not yet heard from the server.
In a case where there is a network partition, the client might not detect it, till the actual TCP timeout that is I think around 30 minutes or more. The requestTimeout setting kicks in before the TCP timeout and then the request is re tried with fresh metadata or errored out depending upon the retry setting. Thanks, Mayuresh On Mon, Aug 29, 2016 at 9:32 AM, Ghosh, Achintya (Contractor) < achintya_gh...@comcast.com> wrote: > Hi Krishna, > Thank you for your response. > > Connections already made but if we increase the request timeout 5 times > let's say request.timeout.ms= 5*60000 , then the number of 'Batch > Expired ' exception is less, so what is the recommended value for ' > request.timeout.ms '. > If we increase more, is there any impact? > > Thanks > Achintya > > -----Original Message----- > From: R Krishna [mailto:krishna...@gmail.com] > Sent: Friday, August 26, 2016 6:17 PM > To: us...@kafka.apache.org > Cc: dev@kafka.apache.org > Subject: Re: Batch Expired > > Are any requests at all making it? That is a pretty big timeout. > > However, I noticed if there is no connections made to broker, you can > still get batch expiry. > > > On Fri, Aug 26, 2016 at 6:32 AM, Ghosh, Achintya (Contractor) < > achintya_gh...@comcast.com> wrote: > > > Hi there, > > > > What is the recommended Producer setting for Producer as I see a lot > > of Batch Expired exception even though I put request.timeout=60000. > > > > Producer settings: > > acks=1 > > retries=3 > > batch.size=16384 > > linger.ms=5 > > buffer.memory=33554432 > > request.timeout.ms=60000 > > timeout.ms=60000 > > > > Thanks > > Achintya > > > > > > -- > Radha Krishna, Proddaturi > 253-234-5657 > -- -Regards, Mayuresh R. Gharat (862) 250-7125
