Following steps are mostly required to Troubleshoot any problem that occurred in production.
As the first step, get the time frame from the user when a particular issue occurred.
Get the logs for that particular time period.
If logs are very large in size, use grep command to filter out errors.
$ grep -o "\w*Exception" error.log | sort -r | uniq -c
It will help to get all the exceptions in error.log sorted in reversed order and give the unique result and with counts.