I just set up a new forwarder on a new Linux server and set it to monitor some files. For some reason, it is sending the data to the indexer twice; I haven't been able to figure out why.
Here's my inputs.conf :
[monitor:/mnt/swift/accounts/*/logfiles/*Queued.log]
ignoreOlderThan = 1d
index = mta-logs_queued
sourcetype = mta-logs_queued
disabled = 0
[monitor:/mnt/swift/accounts/*/logfiles/*Deferred.log]
ignoreOlderThan = 1d
sourcetype = mta-logs_deferred
index = mta-logs_deferred
disabled = 0
[monitor:/mnt/swift/accounts/*/logfiles/*Processed.log]
ignoreOlderThan = 1d
sourcetype = mta-logs_processed
index = mta-logs_processed
disabled = 0
And my props.conf :
[mta-logs_queued]
INDEXED_EXTRACTIONS=csv
NO_BINARY_CHECK=1
SHOULD_LINEMERGE=false
KV_MODE=none
PREAMBLE_REGEX=#
HEADER_FIELD_DELIMITER=,
FIELD_NAMES=Date,Time,EventType,SystemMessageId,CustomMessageId,MailingID,To,From,Size
TIMESTAMP_FIELDS=Date,Time
TIME_FORMAT=%m/%d/%Y %H:%M:%S
TZ=America/Chicago
FIELD_DELIMITER=space
[mta-logs_processed]
INDEXED_EXTRACTIONS=csv
NO_BINARY_CHECK=1
SHOULD_LINEMERGE=false
KV_MODE=none
PREAMBLE_REGEX=#
HEADER_FIELD_DELIMITER=,
FIELD_NAMES=Date,Time,EventType,SystemMessageId,CustomMessageId,MailingID,To,From,Size,ConnectionId,Reason,Message,LocalIp,FailureCode
TIMESTAMP_FIELDS=Date,Time
TIME_FORMAT=%m/%d/%Y %H:%M:%S
TZ=America/Chicago
FIELD_DELIMITER=space
[mta-logs_deferred]
INDEXED_EXTRACTIONS=csv
NO_BINARY_CHECK=1
SHOULD_LINEMERGE=false
KV_MODE=none
PREAMBLE_REGEX=#
HEADER_FIELD_DELIMITER=,
FIELD_NAMES=Date,Time,EventType,SystemMessageId,CustomMessageId,MailingID,To/Domain,From,Size/Count,ConnectionId,Reason,LocalIP
TIMESTAMP_FIELDS=Date,Time
TIME_FORMAT=%m/%d/%Y %H:%M:%S
TZ=America/Chicago
FIELD_DELIMITER=space
Sample raw data:
#Software: SocketLabs, Inc. Hurricane Server.
#Version: 1.0.0.917
#Date: 1/28/2015 00:00:08
#Account: 1000
#Fields: Date Time EventType SystemMessageId CustomMessageId MailingID To From Size
1/28/2015 15:08:02 Queued 6e000000199d02 - - recipient@domain.com sender@domain.com 464
#Software: SocketLabs, Inc. Hurricane Server.
#Version: 1.0.0.917
#Date: 1/28/2015 00:00:09
#Account: 1000
#Fields: Date Time EventType SystemMessageId CustomMessageId MailingID To From Size ConnectionId Reason Message LocalIp FailureCode
1/28/2015 15:08:03 Sent 6e000000199d02 - - recipient@domain.com sender@domain.com 464 6e000000088f08 250+2.5.0+Ok - 10.110.76.77 -
Sample expected event group:
1/28/2015 15:08:02 Queued 6e000000199d02 - - recipient@domain.com sender@domain.com 464
1/28/2015 15:08:03 Sent 6e000000199d02 - - recipient@domain.com sender@domain.com 464 6e000000088f08 250+2.5.0+Ok - 10.110.76.77 -
Sample actual event group:
1/28/2015 15:08:02 Queued 6e000000199d02 - - recipient@domain.com sender@domain.com 464
1/28/2015 15:08:02 Queued 6e000000199d02 - - recipient@domain.com sender@domain.com 464
1/28/2015 15:08:03 Sent 6e000000199d02 - - recipient@domain.com sender@domain.com 464 6e000000088f08 250+2.5.0+Ok - 10.110.76.77 -
1/28/2015 15:08:03 Sent 6e000000199d02 - - recipient@domain.com sender@domain.com 464 6e000000088f08 250+2.5.0+Ok - 10.110.76.77 -
From the splunkd.log :
01-28-2015 14:11:53.156 -0600 INFO TailingProcessor - Parsing configuration stanza: monitor:/mnt/swift/accounts/*/logfiles/*Deferred.log.
01-28-2015 14:11:53.156 -0600 INFO TailingProcessor - Parsing configuration stanza: monitor:/mnt/swift/accounts/*/logfiles/*Processed.log.
01-28-2015 14:11:53.156 -0600 INFO TailingProcessor - Parsing configuration stanza: monitor:/mnt/swift/accounts/*/logfiles/*Queued.log.
01-28-2015 14:11:53.156 -0600 INFO TailingProcessor - Adding watch on path: /mnt/swift/accounts.
01-28-2015 15:08:07.229 -0600 INFO WatchedFile - Resetting fd to re-extract header.
01-28-2015 15:08:07.233 -0600 INFO WatchedFile - Resetting fd to re-extract header.
I also see some lines in this log, Checksum for seekptr didn't match, will re-read entire file='/mnt/swift/accounts/1034/logfiles/20150128Deferred.log'. (but the timestamps don't seem to line up the way the "Resetting fd" ones above do). This is 100% unrelated to the duplicated data. Something's wrong with that particular file.
The inputs are not duplicated anywhere, they live ONLY in /opt/splunkforwarder/etc/system/local/inputs.conf on the forwarder. Ditto with props.conf .
The duplication isn't instantaneous: the first copy of the events comes in, and a couple minutes later, the duplicates show up.
This is the first time I've used the w3c extraction type, so maybe I've configured something wrong?
I'm running Splunk 6.1.2, the forwarder is 6.0.
... View more