[KAFKA-2251] "Connection reset by peer" IOExceptions should not be logged as ERROR

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.9.0.0
• Assignee: Unassigned
• Reporter: Jason Rosenberg
• Created: 04/Jun/15 20:04
• Updated: 15/Jun/15 17:42
• Resolved: 15/Jun/15 07:08

Description

It's normal to see lots of these exceptions logged in the broker logs:

2015-06-04 16:49:30,146 ERROR [kafka-network-thread-27330-1] network.Processor - Closing socket for /1.2.3.4 because of error
java.io.IOException: Connection reset by peer
        at sun.nio.ch.FileDispatcherImpl.read0(Native Method)
        at sun.nio.ch.SocketDispatcher.read(SocketDispatcher.java:39)
        at sun.nio.ch.IOUtil.readIntoNativeBuffer(IOUtil.java:223)
        at sun.nio.ch.IOUtil.read(IOUtil.java:197)
        at sun.nio.ch.SocketChannelImpl.read(SocketChannelImpl.java:379)
        at kafka.utils.Utils$.read(Utils.scala:380)
        at kafka.network.BoundedByteBufferReceive.readFrom(BoundedByteBufferReceive.scala:54)
        at kafka.network.Processor.read(SocketServer.scala:444)
        at kafka.network.Processor.run(SocketServer.scala:340)
        at java.lang.Thread.run(Thread.java:745)

These are routine exceptions, that occur regularly in response to clients going away, etc. The server should not log these as 'ERROR' level, instead they should be probably just 'WARN', and should not log the full stack trace (maybe just the exception message).

The problem is that if we want to alert on actual errors, innocuous errors such as this make it difficult to alert properly, etc.

We are using 0.8.2.1, fwiw