flume-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeff Lord <jl...@cloudera.com>
Subject Re: preserve syslog header in hdfs sink
Date Tue, 01 Apr 2014 16:13:03 GMT
Well you are writing a sequence file (default) Is that what you want?
If you want text use:

hdfs.fileType = datastream

and for the serializer you should be able to just use:

a1.sinks.k1.sink.serializer = header_and_text



On Tue, Apr 1, 2014 at 8:02 AM, Ryan Suarez
<ryan.suarez@sheridancollege.ca>wrote:

>  Thanks for the tip!  I was indeed missing the interceptors.  I've added
> them now but the timestamp and hostname is still not showing up in the hdfs
> log.  Any advice?
>
>
> ------- sample event in HDFS ------
> SEQ
> !org.apache.hadoop.io.LongWritable”org.apache.hadoop.io.BytesWritable������cc�c��I�[��ڳ\�����`���
> �� E � ����Tsu[28432]: pam_unix(su:session): session opened for user root
> by myuser(uid=31043)
>
> ------ same event in syslog ------
> Mar 31 16:18:32 hadoop-t1 su[28432]: pam_unix(su:session): session opened
> for user root by myuser(uid=31043)
>
> ------- flume-conf.properties --------
>
> # Name the components on this agent
> hadoop-t1.sources = r1
> hadoop-t1.sinks = s1
>
> hadoop-t1.channels = mem1
>
> # Describe/configure the source
> hadoop-t1.sources.r1.type = syslogtcp
> hadoop-t1.sources.r1.host = localhost
> hadoop-t1.sources.r1.port = 10005
> hadoop-t1.sources.r1.portHeader = port
> hadoop-t1.sources.r1.interceptors = i1 i2
> hadoop-t1.sources.r1.interceptors.i1.type = timestamp
> hadoop-t1.sources.r1.interceptors.i2.type = host
> hadoop-t1.sources.r1.interceptors.i2.hostHeader = hostname
>
> ##HDFS Sink
> hadoop-t1.sinks.s1.type = hdfs
> hadoop-t1.sinks.s1.hdfs.path = hdfs://
> hadoop-t1.mydomain.org:8020/opt/logs/%{host}/%Y-%m-%d
> hadoop-t1.sinks.s1.hdfs.batchSize = 1
> hadoop-t1.sinks.s1.serializer =
> org.apache.flume.serialization.HeaderAndBodyTextEventSerializer$Builder
> hadoop-t1.sinks.s1.serializer.columns = timestamp hostname
> hadoop-t1.sinks.s1.serializer.format = CSV
> hadoop-t1.sinks.s1.serializer.appendNewline = true
>
> ## MEM  Use a channel which buffers events in memory
>
> hadoop-t1.channels.mem1.type = memory
> hadoop-t1.channels.mem1.capacity = 1000
> hadoop-t1.channels.mem1.transactionCapacity = 100
>
> # Bind the source and sink to the channel
> hadoop-t1.sources.r1.channels = mem1
> hadoop-t1.sinks.s1.channel = mem1
>
>
>
> On 14-03-28 3:37 PM, Jeff Lord wrote:
>
> Do you have the appropriate interceptors configured?
>
>
> On Fri, Mar 28, 2014 at 12:28 PM, Ryan Suarez <
> ryan.suarez@sheridancollege.ca> wrote:
>
>> RTFM indicates I need the following sink properties:
>>
>> ---
>> hadoop-t1.sinks.hdfs1.serializer =
>> org.apache.flume.serialization.HeaderAndBodyTextEventSerializer
>> hadoop-t1.sinks.hdfs1.serializer.columns = timestamp hostname msg
>> hadoop-t1.sinks.hdfs1.serializer.format = CSV
>> hadoop-t1.sinks.hdfs1.serializer.appendNewline = true
>> ---
>>
>> But I'm still not getting timestamp information.  How would I get
>> hostname and timestamp information in the logs?
>>
>>
>> On 14-03-26 3:02 PM, Ryan Suarez wrote:
>>
>>> Greetings,
>>>
>>> I'm running flume that's shipped with Hortonworks HDP2 to feed syslogs
>>> to hdfs.  The problem is the timestamp and hostname of the event is not
>>> logged to hdfs.
>>>
>>> ---
>>> flume@hadoop-t1:~$ hadoop fs -cat
>>> /opt/logs/hadoop-t1/2014-03-26/FlumeData.1395859766307
>>> SEQ!org.apache.hadoop.io.LongWritable"org.apache.hadoop.io.BytesWritable??Ak?i<??G??`D??$hTsu[22209]:
>>> pam_unix(su:session): session opened for user root by someuser(uid=11111)
>>> ---
>>>
>>> How do I configure the sink to add hostname and timestamp info the the
>>> event?
>>>
>>> Here's my flume-conf.properties:
>>>
>>> ---
>>> flume@hadoop-t1:/etc/flume/conf$ cat flume-conf.properties
>>> # Name the components on this agent
>>> hadoop-t1.sources = syslog1
>>> hadoop-t1.sinks = hdfs1
>>> hadoop-t1.channels = mem1
>>>
>>> # Describe/configure the source
>>> hadoop-t1.sources.syslog1.type = syslogtcp
>>> hadoop-t1.sources.syslog1.host = localhost
>>> hadoop-t1.sources.syslog1.port = 10005
>>> hadoop-t1.sources.syslog1.portHeader = port
>>>
>>> ##HDFS Sink
>>> hadoop-t1.sinks.hdfs1.type = hdfs
>>> hadoop-t1.sinks.hdfs1.hdfs.path = hdfs://
>>> hadoop-t1.mydomain.org:8020/opt/logs/%{host}/%Y-%m-%d
>>> hadoop-t1.sinks.hdfs1.hdfs.batchSize = 1
>>>
>>> # Use a channel which buffers events in memory
>>> hadoop-t1.channels.mem1.type = memory
>>> hadoop-t1.channels.mem1.capacity = 1000
>>> hadoop-t1.channels.mem1.transactionCapacity = 100
>>>
>>> # Bind the source and sink to the channel
>>> hadoop-t1.sources.syslog1.channels = mem1
>>> hadoop-t1.sinks.hdfs1.channel = mem1
>>> ---
>>>
>>> ---
>>> flume@hadoop-t1:~$ flume-ng version
>>> Flume 1.4.0.2.0.11.0-1
>>> Source code repository:
>>> https://git-wip-us.apache.org/repos/asf/flume.git
>>> Revision: fcdc3d29a1f249bef653b10b149aea2bc5df892e
>>> Compiled by jenkins on Wed Mar 12 05:11:30 PDT 2014
>>> From source with checksum dea9ae30ce2c27486ae7c76ab7aba020
>>> ---
>>>
>>
>>
>
>

Mime
View raw message