When performing a query that creates a summary report, the associated search.log file shows:
ResultsCollationProcessor - writing remote_event_providers.csv to disk.
Then two hours later reports:
StatsBuffer::read: Row is too large for StatsBuffer, resizing buffer. row_size=77060 needed_space=11536 free_space=153653063
This is soon followed by lots of ~min-by-min output of:
SummaryIndexProcessor - Using tmp_file=/opt/splunk/..../RMD....tmp messages.
What might be happening in that two hour window? We are running Splunk Enterprise 9.1.1 under Linux.