I have a single search that stores many events (~500,000) on the same timestamp.
As I understood, splunk chunks the data to 100,000 and stores the next 100,000 to the next second.
This works fine but, last week splunk stores all the 500,000 to the same timestamp so I can't read the data because I get
"Events may not be returned in sub-second order due to search memory limits configured in limits.conf:[search]:max_rawsize_perchunk. See search.log for more information."
and the search slows down massively.
Does anyone knows why splunk doesn't chunk the results as usual?
Yes, if you have a breakdown in your timestamping, Splunk will default to setting the untimestampable event to the timestamp of the previous event. In this case, you should see splunkd.log logs in index=_internal from your indexers like this:
DateParserVerbose - Failed to parse timestamp. Defaulting to timestamp of previous event ...
The solution is to fix your broken timestamp configuration.