First of all even before reindexing, to configure Splunk to index only recent data, you can use the 2 techniques:
for file monitoring, add the parameter ignoreOlderThan in inputs.conf
It will look at the modtime of the files, example : ignoreOlderThan=7d will index only files touched during the last 7 days. On linux you can couple this with the touch command to change the modtime of a file and trigger the indexing.
see http://docs.splunk.com/Documentation/Splunk/latest/admin/Inputsconf
for WinEventLogs, you can setup the parameter current_only=1 in inputs.conf to exclude the historical logs, and starts only now.
Now that you have setup your inputs to avoid blastering your instance, you can focus on How to force a splunk instance to reindex a file that has already be indexed.
- the radical method is to clean the fishbucket index . That will remove the memory of every files, But it will reindex all.
- on an indexer
splunk clean eventdata -index _fishbucket
on a forwarder by removing the folder $SPLUNK_HOME/var/lib/splunk/fishbucket
or selectivelly forgot a single file from the fishbucket
splunk cmd btprobe -d $SPLUNK_HOME/var/lib/splunk/fishbucket/splunk_private_db --file $FILE --reset
manually reindex each file with the oneshot option,
you also can edit the log file and add a comment on the first line that will force the file to detected as a new file.
./splunk add oneshot "/path/to/my/file.log" -sourcetype mysourcetype
modify the first line of the files to reindex, by default splunk checks the first 256 chars of a file to differentiate them. If you had a simple comment on the first line it wil reindex it
change the crcSalt, create a new input for a new folder, add all the correct sourcetypes, etc...
using a static string that will force a one time reindexing.
crcSalt= REINDEXMEPLEASE
or add the option
crcSalt= <SOURCE>
then move or copy the files to be reindex to the folder, they will be detected as new (because the path will be considered in the crc calculation). (ps the source field will be different of course.)
see http://docs.splunk.com/Documentation/Splunk/latest/admin/Inputsconf
Remark : before reindexing you may want to remove the existing data in splunk to avoid duplicates.
Remark : if you are monitoring windows logs (wineventlog) or are using modular inputs, the counters are not in the fishbucket.
you need to clear the checkpoints files in $SPLUNK_HOME/var/lib/splunk/modinputs/
