Sending JSON Data with HTTP: Events are inconsistent

boverhof — Thu, 07 Feb 2013 18:51:15 GMT

Initially I used the python script to create a log handler to send out JSON formatted log message, but I notice that most of my events had 60 to 70 JSON objects. Some of my events have a single JSON object, which is what I need.

Next I made bash script which uses curl to send each message separately and I got the same thing. I don't understand what's going on here because I'm making a new connection each time ( for each event ) yet sometimes the events are stored as compilations.

$ head -100 test.log | grep Completed
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-287d3871","timestamp":"634908540742893112","message":"SinterConsumer: Completed"}
$ cat test.log | grep Completed | wc -l
197

% cat test.log | grep Completed | while read line ; do curl -u x:$TOKEN "https://api.splunkstorm.com/1/inputs/http?index=XXXXXXXXX&sourcetype=json" -H "Content-type: text/plain" -d "$LINE"; done

The problem here is when I do something like:
spath "instanceID" | search "instanceID"="i-2a7d3873"

I get 6 results ( events ), when in reality there are 197. I pasted event #2 and event #3 so you I can see what I'm talking about.

2 » 2/7/13
12:01:42.000 AM

{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}
{"instanceID":"i-2a7d3873","timestamp":"634908540715990001","message":"SinterConsumer: Completed"}

Show all 67 lines

host=....   Options|  
sourcetype=json   Options|  
source=....   Options

3 » 2/7/13
12:01:35.000 AM

{[-]
instanceID : "i-2a7d3873",
message : "SinterConsumer: Completed",
timestamp : "634908540715990001"
}

Show as raw text

host=....   Options|  
sourcetype=json   Options|  
source=....   Options

Re: Sending JSON Data with HTTP: Events are inconsistent

yannK — Wed, 03 Apr 2013 16:59:27 GMT

json is not a supported sourcetype, it means that splunk will try to guess the timestamp / linebreaking.

please try with one of the 3 supported json sourcetypes, probably the third one.

json_no_timestamp
json_auto_timestamp
json_predefined_timestamp

see http://docs.splunk.com/Documentation/Storm/latest/User/Sourcesandsourcetypes

topic Re: Sending JSON Data with HTTP: Events are inconsistent in Getting Data In

Sending JSON Data with HTTP: Events are inconsistent

Re: Sending JSON Data with HTTP: Events are inconsistent