Splunk Enterprise

Return events based on nested JSON

tsushi
Explorer

Hi,

This is a call log where a participant joined a conference call with audio and video. For example: I want to find all events where rx_packet_loss>1.0 on the video stream.

I don't want any search results based on the element where stream_type=audio in this case.

Been trying to look for answers on my own, but I don't know where to start.

{
        "protocol": "WebRTC", 
        "disconnect_reason": "User initiated disconnect", 
        "media_streams": [
            {
                "rx_packet_loss": 1.5, 
                "stream_type": "video", 
                "tx_packet_loss": 0.0, 
                "id": 585785, 
            }, 
            {
                "rx_packet_loss": 0.0, 
                "stream_type": "audio", 
                "tx_packet_loss": 0.0, 
                "id": 585786, 
            }
        ], 
        "bandwidth": 512, 
        "local_alias": "meet.company.2@domain.com", 
        "call_direction": "in", 
        "remote_alias": "User 1", 
}
0 Karma

acharlieh
Influencer

If you can guarantee order of the fields in media streams, you could use a regular expression to extract the video packet loss as a field, and then filter based on that field:

<> | rex "\"rx_packet_loss\"\s*:\s*(?<video_rx_packetloss>\d+\.\d+)\s*,\s*\"stream_type\"\s*:\s*\"video\"" | where video_rx_packetloss >= 1

Otherwise using some of the multi-valued eval functions, you can combine corresponding stream_type and rx_packet_loss values into a single value ( mvzip), keep only the where stream_type is video ( mvfilter + match), extract out the corresponding rx_packet_loss value ( split + mvindex) and then filter events based on that value.

<> | eval rxloss=mvzip('media_streams{}.stream_type','media_streams{}.rx_packet_loss'),rxloss=mvindex(split(mvfilter(match(rxloss,"^video,")),","),1)| where rxloss >= 1
0 Karma

kamlesh_vaghela
SplunkTrust
SplunkTrust

HI @tsushi,

To work with JSON events generally spath, mvzip, mvexpand & many other multivalued functions will be useful. As part of searching specific events within your provided JSON I think this search will work. Can you please try it?

YOUR_SEARCH | rename media_streams{}.id as id, media_streams{}.rx_packet_loss as rx_packet_loss, media_streams{}.stream_type as stream_type, media_streams{}.tx_packet_loss as tx_packet_loss | eval temp = mvzip(mvzip(mvzip(id,rx_packet_loss),stream_type),tx_packet_loss) | stats count by _time,protocol,disconnect_reason,bandwidth,local_alias,call_direction,remote_alias,temp | eval id=mvindex(split(temp,","),0),rx_packet_loss =mvindex(split(temp,","),1), stream_type=mvindex(split(temp,","),2),tx_packet_loss =mvindex(split(temp,","),3) | where rx_packet_loss>1.0 AND stream_type="video"

Check below link for more informations:

https://docs.splunk.com/Documentation/Splunk/7.0.2/SearchReference/MultivalueEvalFunctions
http://docs.splunk.com/Documentation/Splunk/7.0.2/SearchReference/Spath
http://docs.splunk.com/Documentation/Splunk/7.0.2/SearchReference/Mvexpand

Thanks

0 Karma
Get Updates on the Splunk Community!

Faster Insights with AI, Streamlined Cloud-Native Operations, and More New Lantern ...

Splunk Lantern is a Splunk customer success center that provides practical guidance from Splunk experts on key ...

Splunk Enterprise Security: Your Command Center for PCI DSS Compliance

Every security professional knows the drill. The PCI DSS audit is approaching, and suddenly everyone's asking ...

Developer Spotlight with Guilhem Marchand

From Splunk Engineer to Founder: The Journey Behind TrackMe    After spending over 12 years working full time ...