I am trying to find my average response time of everyday events (not avg of all the events of that day , but the events from 10AM to 1PM) only for last 7 days.
sourcetype="super:access" host=xa20hlf** | eval headers=split(_raw," ") | eval resp_time=mvindex(headers,10) | eval resptime_time_seconds=resp_time*0.001| timechart span=1d eval(round(avg(resptime_time_seconds),2)) as avgTime
Try this...
sourcetype="super:access" host=xa20hlf**
| eval Hour=strftime(_time,"%H")
| where Hour>=10 AND Hour<13
| eval headers=split(_raw," ")
| eval resp_time=mvindex(headers,10)
| eval resptime_time_seconds=resp_time*0.001
| timechart span=1d eval(round(avg(resptime_time_seconds),2)) as avgTim
Hi @zacksoft,
how about this:
sourcetype="super:access" host=xa20hlf** earliest=-7d@d latest=now (date_hour=10 OR date_hour=11 OR date_hour=12 OR date_hour=13) | eval headers=split(_raw," ") | eval resp_time=mvindex(headers,10) | eval resptime_time_seconds=resp_time*0.001| timechart span=1d eval(round(avg(resptime_time_seconds),2)) as avgTime
I wanna change the latest = now to latest = (a fix time of day/week, that I choose like this wednesday 6pm) . Also can we exclude weekends?
When we give span=1d, what exactly is the duration of 1d in clockwise?
Try this...
sourcetype="super:access" host=xa20hlf**
| eval Hour=strftime(_time,"%H")
| where Hour>=10 AND Hour<13
| eval headers=split(_raw," ")
| eval resp_time=mvindex(headers,10)
| eval resptime_time_seconds=resp_time*0.001
| timechart span=1d eval(round(avg(resptime_time_seconds),2)) as avgTim
Dal: In my admittedly limited testing, I found it ~ 8% faster to filter on hour after the timechart. Trim the first eval and the first where, and after timechart add: | where tonumber(strftime(_time,"%H"))>=10 and tonumber(strftime(_time,"%H"))<13
. My test was a simple count on _internal.
@twinspop - you can't filter on hour after the timechart
aggregates to span=1d
... so it would not work unless you change the timechart
to span=1h
and then run it again through timechart
to aggregate it up to the day level... and then you are averaging the averages rather than the actual events. If you have an idea that you think will work, please go ahead and write it up as an answer. The more, the merrier!
Shoot man, you're totally right. My bad.