Hi All, What I'm trying to do is to have a chart with time on x-axis and percentages by ResponseStatus on y-axis. To do that I come up with the below Splunk search query: match some http requests
| fields _time,ResponseStatus,RequestName
| eval Date=strftime(_time, "%m/%d/%Y")
| eval ResponseStatus=if(isnull(ResponseStatus), 504, ResponseStatus)
| eventstats count as "totalCount" by Date
| eventstats count as "codeCount" by Date,ResponseStatus
| eval percent=round((codecount/totalCount)*100)
| chart values(percent) by Date,ResponseStatus But it is hitting the disk usage limit (500MB - which I can't increase) for a 10 days interval. And I'd like to be able to have this on a 3/4 months interval. What I have noticed is that If I only run the match part of the query, I get all the events without hitting any disk limit, which makes me think the problem is with the counting and group by part of the query. My guess is that Splunk is making the computation by keeping in-memory (or, trying to do so and eventually swapping to disk) the full event message even if I specified the useful fields via the fields command. Is there any way to either effectively have Splunk ignore all the remaining part of the message or obtain the same result via a different path? Thanks a lot!
... View more