You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@flume.apache.org by Josh Myers <jo...@mydrivesolutions.com> on 2013/06/17 15:46:47 UTC
Flume events rolling file too regularly
Hi guys,
We are sending JSON events from our pipeline into a flume http source.
We have written a custom multiplexer and sink serializer. The events are
being routed into the correct channels and consumed OK by the sinks. The
custom serializer takes a JSON event and outputs a csv. Files are being
written to s3 ( using s3n as hdfs ) but rather than appending the
written csv file, each event seems to be generating it own csv. The
output is what I would expect using rollCount 1, however we do
occasionally get several events ( maybe 4 ) written per csv. Please see
below for config.
Ideally we want to use rollInterval of 24 hours, to generate a new .csv
file every 24 hours, but have events pretty quickly flushed to the csv
file after being sent. So one csv' per day that is consistently appended
with whatever events we throw in. We found however that with a
rollInterval of 24 hours the events weren't being flushed often enough...
Any help would be hugely appreciated!
Thanks.
Josh
## Sources ###################################################
agent.sources = http
agent.sources.http.type = http
agent.sources.http.bind = 0.0.0.0
agent.sources.http.port = 4444
agent.sources.http.channels = cappucino_s3_aggregate_profile_channel
default_s3_channel cappucino_s3_trip_summary_channel
## Interceptors #################################################
agent.sources.http.interceptors = itime ihost
agent.sources.http.interceptors.itime.type = timestamp
agent.sources.http.interceptors.ihost.type = host
agent.sources.http.interceptors.ihost.useIP = false
agent.sources.http.interceptors.ihost.preserveExisting= false
agent.sources.http.interceptors.ihost.hostHeader = hostname
## Multiplex Channels Mapping ######################################
agent.sources.http.selector.type =
com.mydrivesolutions.flume.serialization.PipelineEventsChannelSelector
agent.sources.http.selector.default = default_s3_channel
## Channels ########################################################
agent.channels = cappucino_s3_aggregate_profile_channel
cappucino_s3_trip_summary_channel default_s3_channel
agent.channels.cappucino_s3_aggregate_profile_channel.type = file
agent.channels.cappucino_s3_aggregate_profile_channel.capacity = 10000000
agent.channels.cappucino_s3_aggregate_profile_channel.checkpointDir =
/mnt/flume/cappucino_s3_aggregate_profile/checkpoint
agent.channels.cappucino_s3_aggregate_profile_channel.dataDirs =
/mnt/flume/cappucino_s3_aggregate_profile/data
agent.channels.cappucino_s3_trip_summary_channel.type = file
agent.channels.cappucino_s3_trip_summary_channel.capacity = 10000000
agent.channels.cappucino_s3_trip_summary_channel.checkpointDir =
/mnt/flume/cappucino_s3_trip_summary/checkpoint
agent.channels.cappucino_s3_trip_summary_channel.dataDirs =
/mnt/flume/cappucino_s3_trip_summary/data
## Sinks ###########################################################
agent.sinks = cappucino_s3_aggregate_profile_sink1
cappucino_s3_aggregate_profile_sink2 cappucino_s3_trip_summary_sink1
cappucino_s3_trip_summary_sink2
## Serialize json events from the pipeline and write csv to HDFS (We are
using s3 native FS as HDFS)
###############################################################################
## Capuccino_s3_aggregate_profile Sinks
#################################################
agent.sinks.cappucino_s3_aggregate_profile_sink1.type = hdfs
agent.sinks.cappucino_s3_aggregate_profile_sink1.channel =
cappucino_s3_aggregate_profile_channel
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.path =
s3n://mydrive-cappucino-reports/driver-profiles/%Y-%m-%d
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.fileType = DataStream
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.writeFormat = Text
agent.sinks.cappucino_s3_aggregate_profile_sink1.serializer =
com.mydrivesolutions.flume.serialization.HeaderAndBodyTextEventSerializer$Builder
agent.sinks.cappucino_s3_aggregate_profile_sink1.serializer.columns =
log_type reporting_bucket subscription_id
agent.sinks.cappucino_s3_aggregate_profile_sink1.serializer.format =
DriverProfile
agent.sinks.cappucino_s3_aggregate_profile_sink1.serializer.delimiter = ,
agent.sinks.cappucino_s3_aggregate_profile_sink1.serializer.appendNewline =
false
agent.sinks.cappucino_s3_aggregate_profile_sink1.serializer.distanceMeasure
= MILES
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.maxOpenFiles = 5000
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.rollInterval = 20400
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.callTimeout = 60000
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.fileSuffix = .csv
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.inUseSuffix = .tmp
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.filePrefix =
DriverProfile.%y-%m-%d.%H.%M
agent.sinks.cappucino_s3_aggregate_profile_sink1.hdfs.timeZone = UTC
agent.sinks.cappucino_s3_aggregate_profile_sink2.type = hdfs
agent.sinks.cappucino_s3_aggregate_profile_sink2.channel =
cappucino_s3_aggregate_profile_channel
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.path =
s3n://mydrive-cappucino-reports/driver-profiles/%Y-%m-%d
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.fileType = DataStream
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.writeFormat = Text
agent.sinks.cappucino_s3_aggregate_profile_sink2.serializer =
com.mydrivesolutions.flume.serialization.HeaderAndBodyTextEventSerializer$Builder
agent.sinks.cappucino_s3_aggregate_profile_sink2.serializer.columns =
log_type reporting_bucket subscription_id
agent.sinks.cappucino_s3_aggregate_profile_sink2.serializer.format =
DriverProfile
agent.sinks.cappucino_s3_aggregate_profile_sink2.serializer.delimiter = ,
agent.sinks.cappucino_s3_aggregate_profile_sink2.serializer.appendNewline =
false
agent.sinks.cappucino_s3_aggregate_profile_sink2.serializer.distanceMeasure
= MILES
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.maxOpenFiles = 5000
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.rollInterval = 20400
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.callTimeout = 60000
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.fileSuffix = .csv
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.inUseSuffix = .tmp
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.filePrefix =
DriverProfile.%y-%m-%d.%H.%M
agent.sinks.cappucino_s3_aggregate_profile_sink2.hdfs.timeZone = UTC
## Confused_s3_trip_summary Sinks
#################################################
agent.sinks.cappucino_s3_trip_summary_sink1.type = hdfs
agent.sinks.cappucino_s3_trip_summary_sink1.channel =
cappucino_s3_trip_summary_channel
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.path =
s3n://mydrive-cappucino-reports/trip-summaries/%Y-%m-%d
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.fileType = DataStream
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.writeFormat = Text
agent.sinks.cappucino_s3_trip_summary_sink1.serializer =
com.mydrivesolutions.flume.serialization.HeaderAndBodyTextEventSerializer$Builder
agent.sinks.cappucino_s3_trip_summary_sink1.serializer.columns =
log_type reporting_bucket subscription_id
agent.sinks.cappucino_s3_trip_summary_sink1.serializer.format = BodyCSV
agent.sinks.cappucino_s3_trip_summary_sink1.serializer.delimiter = ,
agent.sinks.cappucino_s3_trip_summary_sink1.serializer.appendNewline = false
agent.sinks.cappucino_s3_trip_summary_sink1.serializer.distanceMeasure =
MILES
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.maxOpenFiles = 5000
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.rollInterval = 20400
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.callTimeout = 60000
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.fileSuffix = .csv
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.inUseSuffix = .tmp
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.filePrefix =
TripSummary.%y-%m-%d.%H.%M
agent.sinks.cappucino_s3_trip_summary_sink1.hdfs.timeZone = UTC
agent.sinks.cappucino_s3_trip_summary_sink2.type = hdfs
agent.sinks.cappucino_s3_trip_summary_sink2.channel =
cappucino_s3_trip_summary_channel
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.path =
s3n://mydrive-cappucino-reports/trip-summaries/%Y-%m-%d
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.fileType = DataStream
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.writeFormat = Text
agent.sinks.cappucino_s3_trip_summary_sink2.serializer =
com.mydrivesolutions.flume.serialization.HeaderAndBodyTextEventSerializer$Builder
agent.sinks.cappucino_s3_trip_summary_sink2.serializer.columns =
log_type reporting_bucket subscription_id
agent.sinks.cappucino_s3_trip_summary_sink2.serializer.format = BodyCSV
agent.sinks.cappucino_s3_trip_summary_sink2.serializer.delimiter = ,
agent.sinks.cappucino_s3_trip_summary_sink2.serializer.appendNewline = false
agent.sinks.cappucino_s3_trip_summary_sink2.serializer.distanceMeasure =
MILES
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.maxOpenFiles = 5000
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.rollInterval = 20400
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.callTimeout = 60000
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.fileSuffix = .csv
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.inUseSuffix = .tmp
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.filePrefix =
TripSummary.%y-%m-%d.%H.%M
agent.sinks.cappucino_s3_trip_summary_sink2.hdfs.timeZone = UTC
## SinkGroups ###########################################################
agent.sinkgroups = cappucino_s3_aggregate_profile_sinkgroup
cappucino_s3_trip_summary_sinkgroup
## Confused_s3_aggregate_profile Failover SinkGroup
##########################################
agent.sinkgroups.cappucino_s3_aggregate_profile_sinkgroup.sinks =
cappucino_s3_aggregate_profile_sink1 cappucino_s3_aggregate_profile_sink2
agent.sinkgroups.cappucino_s3_aggregate_profile_sinkgroup.processor.type
= failover
agent.sinkgroups.cappucino_s3_aggregate_profile_sinkgroup.processor.priority.cappucino_s3_aggregate_profile_sink1
= 10
agent.sinkgroups.cappucino_s3_aggregate_profile_sinkgroup.processor.priority.cappucino_s3_aggregate_profile_sink2
= 5
agent.sinkgroups.cappucino_s3_aggregate_profile_sinkgroup.processor.maxpenalty
= 30000
## Confused_s3_trip_summary Failover SinkGroup
##########################################
agent.sinkgroups.cappucino_s3_trip_summary_sinkgroup.sinks =
cappucino_s3_trip_summary_sink1 cappucino_s3_trip_summary_sink2
agent.sinkgroups.cappucino_s3_trip_summary_sinkgroup.processor.type =
failover
agent.sinkgroups.cappucino_s3_trip_summary_sinkgroup.processor.priority.cappucino_s3_trip_summary_sink1
= 10
agent.sinkgroups.cappucino_s3_trip_summary_sinkgroup.processor.priority.cappucino_s3_trip_summary_sink2
= 5
agent.sinkgroups.cappucino_s3_trip_summary_sinkgroup.processor.maxpenalty =
30000
--
www.mydrivesolutions.com
This email and any attachments is private and confidential. If you have
received this message in error please remove it from your systems and
notify the author.
MyDrive Solutions Limited is registered in England and Wales, No 07330334.
Registered office: Surrey Technology Centre, 40 Occam Road, Guildford GU2
7YG, UK
RE: Flume events rolling file too regularly
Posted by Josh Myers <jo...@mydrivesolutions.com>.
Thanks for that Paul, we have rollInterval set at 21600 and have
experienced the same problem when setting rollSize 0, rollCount 0 in order
to disable rolling on defaults. Any other ideas? We want one csv file per
day that is continuously flushed to..
Thanks
Josh
On 17 Jun 2013 17:16, "Paul Chavez" <pc...@verticalsearchworks.com> wrote:
> There are three file roll defaults on the HDFS sink, rollInterval,
> rollSize and rollCount. Their defaults are 30s, 1024B and 10 ecents,
> respectively. You need to set each one as desired or disable them.
>
> My mail client chopped up your config, but I did a search and didn't see
> any of those properties set. I would start there.
>
> Hope that helps,
> Paul Chavez
>
> -----Original Message-----
> From: Josh Myers [mailto:josh.myers@mydrivesolutions.com]
> Sent: Monday, June 17, 2013 6:47 AM
> To: user@flume.apache.org
> Subject: Flume events rolling file too regularly
>
> Hi guys,
>
> We are sending JSON events from our pipeline into a flume http source.
> We have written a custom multiplexer and sink serializer. The events are
> being routed into the correct channels and consumed OK by the sinks. The
> custom serializer takes a JSON event and outputs a csv. Files are being
> written to s3 ( using s3n as hdfs ) but rather than appending the written
> csv file, each event seems to be generating it own csv. The output is what
> I would expect using rollCount 1, however we do occasionally get several
> events ( maybe 4 ) written per csv. Please see below for config.
>
> Ideally we want to use rollInterval of 24 hours, to generate a new .csv
> file every 24 hours, but have events pretty quickly flushed to the csv file
> after being sent. So one csv' per day that is consistently appended with
> whatever events we throw in. We found however that with a rollInterval of
> 24 hours the events weren't being flushed often enough...
>
> Any help would be hugely appreciated!
>
> Thanks.
>
>
> Josh
>
--
www.mydrivesolutions.com
This email and any attachments is private and confidential. If you have
received this message in error please remove it from your systems and
notify the author.
MyDrive Solutions Limited is registered in England and Wales, No 07330334.
Registered office: Surrey Technology Centre, 40 Occam Road, Guildford GU2
7YG, UK
RE: Flume events rolling file too regularly
Posted by Paul Chavez <pc...@verticalsearchworks.com>.
There are three file roll defaults on the HDFS sink, rollInterval, rollSize and rollCount. Their defaults are 30s, 1024B and 10 ecents, respectively. You need to set each one as desired or disable them.
My mail client chopped up your config, but I did a search and didn't see any of those properties set. I would start there.
Hope that helps,
Paul Chavez
-----Original Message-----
From: Josh Myers [mailto:josh.myers@mydrivesolutions.com]
Sent: Monday, June 17, 2013 6:47 AM
To: user@flume.apache.org
Subject: Flume events rolling file too regularly
Hi guys,
We are sending JSON events from our pipeline into a flume http source.
We have written a custom multiplexer and sink serializer. The events are being routed into the correct channels and consumed OK by the sinks. The custom serializer takes a JSON event and outputs a csv. Files are being written to s3 ( using s3n as hdfs ) but rather than appending the written csv file, each event seems to be generating it own csv. The output is what I would expect using rollCount 1, however we do occasionally get several events ( maybe 4 ) written per csv. Please see below for config.
Ideally we want to use rollInterval of 24 hours, to generate a new .csv file every 24 hours, but have events pretty quickly flushed to the csv file after being sent. So one csv' per day that is consistently appended with whatever events we throw in. We found however that with a rollInterval of 24 hours the events weren't being flushed often enough...
Any help would be hugely appreciated!
Thanks.
Josh