I have two separate searches that I want to group into one. When I use appendcols
I get wrong counts for the search encapsulated within appendcols
. Can someone clue me into what I'm doing wrong?
In the search below, "Provisioned Org" returns an incorrect count, than when I run it on its own.
sourcetype=logs statusCode=400 "Org failure" earliest=-1mon@mon latest=@mon| timechart span=1d count as FAILED|appendcols [search sourcetype=logs "Provisioned org" earliest=-1mon@mon latest=@mon | timechart span=1d count as SUCCESSFUL]
Appendcols will not be able to correlate too many events. Considering the fact that your are roughly trying to aggregate one months of data for successful and unsuccessful events, there might be more events than what can be handled based on your Splunk configurations (hardware and limits). You would notice two things search running too slow and older dates returning 0 counts.
1) You can either run appendcols for relatively shorter period of time like a week or single day.
2) If stausCode field or any other field for correlation is present for both successful and failed events then use stats/timechart command instead of any other correlation techniques like append, appendcols or join. Assuming 200 is successful and rest all (including 400) are failed.
sourcetype=logs "Org failure" OR "Provisioned org" statusCode=* earliest=-1mon@mon latest=@mon| timechart span=1d count(eval(statusCode=200)) as SUCCESS, count(eval(statusCode!=200)) as FAILED
Appendcols will not be able to correlate too many events. Considering the fact that your are roughly trying to aggregate one months of data for successful and unsuccessful events, there might be more events than what can be handled based on your Splunk configurations (hardware and limits). You would notice two things search running too slow and older dates returning 0 counts.
1) You can either run appendcols for relatively shorter period of time like a week or single day.
2) If stausCode field or any other field for correlation is present for both successful and failed events then use stats/timechart command instead of any other correlation techniques like append, appendcols or join. Assuming 200 is successful and rest all (including 400) are failed.
sourcetype=logs "Org failure" OR "Provisioned org" statusCode=* earliest=-1mon@mon latest=@mon| timechart span=1d count(eval(statusCode=200)) as SUCCESS, count(eval(statusCode!=200)) as FAILED
Thank you. Yes, I did notice both conditions you mentioned - slow queries and 0 counts.