flume报错 Configured capacity is 100000000 but the checkpoint file capacity is 10

xlid

　　最近flume运行不是很稳定，本次由于hadoop不能写入，导致flume报错，Configured capacity is 100000000 but the  checkpoint file capacity is 1000000，重启flume后问题仍然存在。
　　1，详细报错如下：

　　22 Jan 2013 11:07:42,568 INFO  [pool-7-thread-1] (org.apache.avro.ipc.NettyServer$NettyServerAvroHandler.channelClosed:209)  - Connection to /10.4.203.176:60322 disconnected.
22 Jan 2013 11:07:44,617 ERROR [SinkRunner-PollingRunner-DefaultSinkProcessor] (org.apache.flume.SinkRunner$PollingRunner.run:160)  - Unable to deliver event. Exception follows.
java.lang.IllegalStateException: Channel closed [channel=file_chan_1]. Due to java.lang.IllegalStateException: Configured capacity is 100000000 but the  checkpoint file capacity is 1000000. See FileChannel documentation on how to change a channels capacity.
        at org.apache.flume.channel.file.FileChannel.createTransaction(FileChannel.java:321)
        at org.apache.flume.channel.BasicChannelSemantics.getTransaction(BasicChannelSemantics.java:122)
        at org.apache.flume.sink.hdfs.HDFSEventSink.process(HDFSEventSink.java:385)
        at org.apache.flume.sink.DefaultSinkProcessor.process(DefaultSinkProcessor.java:68)
        at org.apache.flume.SinkRunner$PollingRunner.run(SinkRunner.java:147)
        at java.lang.Thread.run(Thread.java:722)
Caused by: java.lang.IllegalStateException: Configured capacity is 100000000 but the  checkpoint file capacity is 1000000. See FileChannel documentation on how to change a channels capacity.
        at org.apache.flume.channel.file.EventQueueBackingStoreFile.<init>(EventQueueBackingStoreFile.java:80)
        at org.apache.flume.channel.file.EventQueueBackingStoreFileV3.<init>(EventQueueBackingStoreFileV3.java:42)
        at org.apache.flume.channel.file.EventQueueBackingStoreFactory.get(EventQueueBackingStoreFactory.java:67)
        at org.apache.flume.channel.file.EventQueueBackingStoreFactory.get(EventQueueBackingStoreFactory.java:36)
        at org.apache.flume.channel.file.Log.replay(Log.java:339)
        at org.apache.flume.channel.file.FileChannel.start(FileChannel.java:271)
        at org.apache.flume.lifecycle.LifecycleSupervisor$MonitorRunnable.run(LifecycleSupervisor.java:236)
        at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:471)
        at java.util.concurrent.FutureTask$Sync.innerRunAndReset(FutureTask.java:351)
        at java.util.concurrent.FutureTask.runAndReset(FutureTask.java:178)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$301(ScheduledThreadPoolExecutor.java:178)
        at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1110)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:603)
        ... 1 more
        
2，故障原因：
    FileChannel使用了固定大小的checkpoint file，修改channel的能力，简单的方法如下
    1，关闭agent
    2, 删除或者备份checkpoint目录
    3，重启flume agent（如果channel中有大量文件，会导致全面的延迟，建议先关闭数据源，等把file channel中的数据全部写入sink后，等上2分钟，等数据文件删除后，再重启channel
    具体详见参考资料这段话，这段意思未完全读明白，通过解决方法靠谱
        
The FileChannel actually uses a fixed size checkpoint file -- so it is not possible to set
it to unlimited size (the checkpoint file is mmap-ed to a fixed size buffer). To change the
capacity of the channel, the easiest way off the top of my head is:

* Shutdown the agent.
* Delete all files in the file channel's checkpoint directory. (not the data directories.
Also you might want to move them out, rather than delete to be safe)
* Change your configuration to increase the capacity of the channel.
* Restart the agent - this will cause full replay, so the agent might take sometime to start
up if there are a lot of events in the channel (to avoid this - shutdown the source before
shutting the agent down - so the sink can drain out the channel completely, wait for about
1-2 mins after the channel is empty so that the data files get deleted (this happens only
immediately after a checkpoint - you can verify this by making sure each data dir has only
2 files each), since all events have been sent out - so during restart the channel will be
quite empty, with very little to replay).


参考资料
http://mail-archives.apache.org/mod_mbox/flume-user/201211.mbox/%3CDA66E7657BFBD949829F3C98D73A54CF18E23C@RHV-EXRDA-S11.corp.ebay.com%3E