What are the limitations regarding lookups?

[DerBastler]

Hey there. This isn't a specific code question but rather a more general question regarding limitations of lookups.
From programming in other languages I experienced it now and then that a file was "locked" by a program or user and it could not be written to by another program or user.

I am wondering if |outputlookup xyz.csv has a similar limitation.
let's say if there are dozens of alerts which also have an outputlookup command and trying to write at the same second, or if there are multiple users using a dashboard which includes an outputlookup - will the csv be able to handle it? Will they queue up? or will some attempts to write to the file be just ignored? If there are limitations, how do they differ when using a KV Store instead of a csv file?

Thank you!

[Ayn · Accepted solution]

Yes, this is a potential issue with using CSV files for lookups (one I've been bitten by myself and had to spend quite some time before discovering what the problem was), race conditions can occur when multiple users want to lock the same file for writing. It is one good reason to use KV store based lookups instead. A comparison between CSV based and KV store based lookups is available on this page: http://docs.splunk.com/Documentation/Splunk/7.1.1/Knowledge/DefineaKVStorelookupinSplunkWeb#KV_Store...

[DerBastler]

Alrighty, thank you!