Getting Data In

Script abruptly stopped working

Branden
Builder

I have a script that runs in an app on my forwarders every 12 hours. Or at least that's what it was doing until it abruptly stopped working on all forwarders on September 30th.

The script is deployed using the deployment server, so all the forwarders have identical configurations.

The inputs.conf file has five scripts configured and all work except for this one. Here is its inputs.conf entry:

[script://splunk/etc/apps/all/bin/chknmon.sh]
interval = 43200 # Run every 12 hours
sourcetype = nmonchk
source = script://./bin/chknmon.sh

The path to the chknmon.sh script is correct:

[/splunk/etc/apps/all/default]# ls -las /splunk/etc/apps/all/bin/chknmon.sh
4 -rwxr-xr--    1 xyz abc 96 Oct 04 14:13 /splunk/etc/apps/all/bin/chknmon.sh

Splunk runs as root, and the script is owned by root.

Running the script manually produces correct output.

Other scripts in the same app (and same paths/permissions) run just fine. This one was no exception until it abruptly stopped working a few days ago. Admittedly, I've been toying around with Splunk a bit, but I have no idea what I could have done that could have affected just this one script.

I've tried bouncing both the indexer and the forwarders, but that didn't help.

I'm running 4.1.4 on the forwarders and 4.1.5 on the indexer.

Thanks!

0 Karma
1 Solution

Lowell
Super Champion

Did it stop on Sept 30, or after Sept 30th? I'm just thinking that you could have a timestamp parsing / configuration issue.

There are two things that could cause a date passing issues starting Oct 1st. First, this is the first time in 2010 that the month is 2 digits (depending on whether your data format uses "9" or "09" for the month of September). Anything that assumed a single digit could have issues with this (for example, if you use splunk's punct pattern anywhere in a search or eventtype). Also, Oct 1st, is the first time time the month is the same as the 2 digit year; so if your timestamps format is in any way ambiguous, and splunk is guessing about your timeformat, then this could cause some confusion. You may want to do a search across "All Time" and see if your events were simply given the wrong timestamp.

The bottom line: If your props.conf entry for [nmonchk] doesn't have an explicit TIME_FORMAT entry, then I would suggest adding one.

View solution in original post

Lowell
Super Champion

Did it stop on Sept 30, or after Sept 30th? I'm just thinking that you could have a timestamp parsing / configuration issue.

There are two things that could cause a date passing issues starting Oct 1st. First, this is the first time in 2010 that the month is 2 digits (depending on whether your data format uses "9" or "09" for the month of September). Anything that assumed a single digit could have issues with this (for example, if you use splunk's punct pattern anywhere in a search or eventtype). Also, Oct 1st, is the first time time the month is the same as the 2 digit year; so if your timestamps format is in any way ambiguous, and splunk is guessing about your timeformat, then this could cause some confusion. You may want to do a search across "All Time" and see if your events were simply given the wrong timestamp.

The bottom line: If your props.conf entry for [nmonchk] doesn't have an explicit TIME_FORMAT entry, then I would suggest adding one.

Branden
Builder

Thanks! I managed to get the TIME_FORMAT right, but now that I read your comment I think I'm better off with DATETIME_CONFIG=CURRENT.

0 Karma

Lowell
Super Champion

Actually, if you don't want splunk to interpret any date times, set DATETIME_CONFIG=CURRENT for your sourcetype in props.conf. If you need help with a TIME_FORMAT value, simply post a couple example timestamps and someone will give you a hand; based on your 02/10/10 example, I would guess you want TIME_FORMAT = %d/%m/%y %H:%M:%S, but that's just a guess. You probably don't need to mess with TIME_PREFIX, unless you have multiple timestamps in your event. If you set DATETIME_CONFIG=CURRENT then you don't need either of the TIME_* settings.

0 Karma

Branden
Builder

Actually I think I know where to go from here. The TIME_PREFIX regular expression is a bit tricky for this one, but I'll get it with trial and error.
Thanks again!

0 Karma

Branden
Builder

Wow....
Stopped at 09/30/2010 at 12:01 AM.

I took your suggestion and did an "all time" search... and I'm finding entries from 02/10/10... which is impossible because I didn't have this script running in February. Nice catch.

So I guess my question now is: what do I do with TIME_FORMAT? I just want the index time to be the time it invokes the script...

Thanks!

0 Karma
Get Updates on the Splunk Community!

What's new in Splunk Cloud Platform 9.1.2312?

Hi Splunky people! We are excited to share the newest updates in Splunk Cloud Platform 9.1.2312! Analysts can ...

What’s New in Splunk Security Essentials 3.8.0?

Splunk Security Essentials (SSE) is an app that can amplify the power of your existing Splunk Cloud Platform, ...

Let’s Get You Certified – Vegas-Style at .conf24

Are you ready to level up your Splunk game? Then, let’s get you certified live at .conf24 – our annual user ...