Splunk Search

How to only display rows in table when one field changes?

bensec01
Explorer

Hey folks,

I have a hard time believing this hasn't come up before, but I didn't find the right kinds of questions/answers when I went a-searching.

I have an inventory of hosts (specifically, data from a VMware DCN) that gives me periodic information about my virtual hosts. Super awesome, two thumbs up. I'm just starting to really dig into it and get more useful bits out.

The bit I'm working on now is tracking a VMs movement through our clusters. Ie, if a VM is vmotioned to another host, I want to see that. So, I have a very very simple search that spits out a table:

tag=virtualmachine tag=inventory tag=virtualization vm_name="some-VM-name" changeSet.summary.runtime.host.name=*
| table _time, hypervisor_name

This spits out:

2017-08-31 12:00:00 dc1-host-63
2017-08-31 11:00:00 dc1-host-63
2017-08-31 10:00:00 dc1-host-63
2017-08-31 09:00:00 dc1-host-41
2017-08-31 08:00:00 dc1-host-09
2017-08-31 07:00:00 dc1-host-09
2017-08-31 06:00:00 dc1-host-63
2017-08-31 05:00:00 dc1-host-63

etc, etc.

What I'd really like to see, is just when the second column changes - I don't need to know that the VM was still on dc1-host-63 at 11:00 today. Instead, I want to see the first event when that second column has changed from the previous value (in this example, when the inventory run first sees the VM on a new host):

2017-08-31 10:00:00 dc1-host-63
2017-08-31 09:00:00 dc1-host-41
2017-08-31 08:00:00 dc1-host-09
2017-08-31 05:00:00 dc1-host-63

I know dedup, but I can't dedup the time because I want to keep that timestamp intact where the second column changes. And I can't just dedup the second column, because then when a VM moved back to a previous host, I'd lose a row. I have the feeling dedup isn't going to be the right sauce for this one, but I haven't found the magic yet.

Any help would be very much appreciated!

Chris

0 Karma
1 Solution

DalJeanis
Legend

try this

tag=virtualmachine tag=inventory tag=virtualization vm_name="some-VM-name" changeSet.summary.runtime.host.name=*
| table _time, hypervisor_name
| sort 0 _time
| streamstats current=f window=2 last(hypervisor_name) as prior_name
| where hypervisor_name!=prior_name OR isnull(prior_name)
| fields - prior_name

View solution in original post

DalJeanis
Legend

try this

tag=virtualmachine tag=inventory tag=virtualization vm_name="some-VM-name" changeSet.summary.runtime.host.name=*
| table _time, hypervisor_name
| sort 0 _time
| streamstats current=f window=2 last(hypervisor_name) as prior_name
| where hypervisor_name!=prior_name OR isnull(prior_name)
| fields - prior_name

bensec01
Explorer

Awesome. I had to add a space after the hyphen in fields, but it works perfectly. I thought streamstats might be a possibility, but I'm pretty green with that one...

Thank you so much!

Got questions? Get answers!

Join the Splunk Community Slack to learn, troubleshoot, and make connections with fellow Splunk practitioners in real time!

Meet up IRL or virtually!

Join Splunk User Groups to connect and learn in-person by region or remotely by topic or industry.

Get Updates on the Splunk Community!

ATTENTION: We’re Moving! (AGAIN!)

The Splunk Community Slack is undergoing a system migration to keep our workspace secure and ...

Deep Dive: Optimizing Telemetry Pipelines in Splunk Observability Cloud

In this session, we will peel back the layers of Splunk Observability Cloud’s cost-optimization features. ...

Announcing Modern Navigation: A New Era of Splunk User Experience

We are excited to introduce the Modern Navigation feature in the Splunk Platform, available to both cloud and ...