Scripted Inputs: how do I persist state between invocations of script?

[maratc]

I have a script that pulls events from my REST API for Splunk to index. My script runs on schedule.

I want to only pull new events, to prevent duplication and unnecessary traffic. My events have incrementing IDs.

To pull new events I need my script to remember what was the ID of the last pulled event, i.e. my script needs to persist state between runs. If Splunk instance restarts, I too wouldn't like to bring all the events from the beginning.

What are my options here? I would like not to read last ID by issuing query to Splunk.

Thanks!

[jkat54]

You can also run a search with curl or other http tools to get the sourcetype=mysource | stats first(ID) by sourcetype, then use the results of this search in your script so that Splunk is your "database" / "config file".

If you do the filesystem file as suggested above, I recommend something like appName/bin/.deltaFile

the . in front of the filename will make it hidden on linux systems. Heres some unoptimized code for creating/reading/updating a delta file which contains the last date of execution,

def getDeltaDate(datapath):
    try:
        if os.path.exists(datapath + '/.delta_date'):
            #open delta file and return date from file
            with open(datapath + '/.delta_date','r') as deltafile:
                for lastrundate in deltafile:
                    return lastrundate
                deltafile.close()
            #write new date to file, overwriting the original
            with open(datapath + '/.delta_date','w') as deltafile:
                deltafile.write(str(date.today()))
            deltafile.close()
        else:
            #write new date to file, original shouldnt exist at this point but will overwrite if so
            with open(datapath + '/.delta_date','w') as deltafile:
                deltafile.write(str(date.today()))
            deltafile.close()
            firstDate = "1970-01-01"
            return firstDate
    except OSError as e:
        logger.critical('Function: getDeltaDate failed due to the following error(s): ' + str(e))
        print('Function: getDeltaDate failed due to the following error(s): ' + str(e))

[javiergn]

Wouldn't a local file where you store that sort of information work for you?

[maratc]

I guess it would; is my script permitted to write to files, and is that a reasonable expectation that the files will stay there between restarts?