topic Re: Non-Padded Milliseconds on Timestamp in Getting Data In

Non-Padded Milliseconds on Timestamp

darrenfuller — Fri, 18 Oct 2019 18:44:59 GMT

Hello..

Splunk 7.0.5, I have a data source as follows, which has 1, 2 or 3 digit values for millisecond. :

Tue Oct 08 2019 14:47:33 tid="d83af63f5acd9c510bd440d" object="EnterpriseSalesInformation" rule_type="request" domain="SOAPvs" domain="SOAPvs" trace_id="30fb2039-9910-4bd5-b012-3fe63dd874af" time="2019-10-08T14:47:33:4-04:00" 
Tue Oct 08 2019 14:47:29 tid="d83af63f5acd9c180b220fd" object="EnterpriseSalesInformation" rule_type="request" domain="SOAPvs" domain="SOAPvs" trace_id="24c792c0-80ea-476b-95b2-3fe63dd8876c" time="2019-10-08T14:47:29:91-04:00" 
Tue Oct 08 2019 14:47:29 tid="d83af63f5acd9c180bd203d" object="EnterpriseSalesInformation" rule_type="request" domain="SOAPvs" domain="SOAPvs" trace_id="7446787d-76b8-4eca-924f-3fe63dd8c9ed" time="2019-10-08T14:47:29:181-04:00"

Which i am trying to process using this props.conf:

[teshting]
disabled = false
DATETIME_CONFIG = /etc/apps/teshtapp/local/customdatetime.xml
TIME_PREFIX = time="
# TIME_FORMAT = %Y-%m-%dT%H:%M:%S:%3Q%z
# TIME_FORMAT = %Y-%m-%dT%H:%M:%S:%3N%z
MAX_TIMESTAMP_LOOKAHEAD = 40
LINE_BREAKER = ([\r\n]+)
SHOULD_LINEMERGE = false
TRUNCATE = 1000

( You can see from my commented lines, i tried to deal with this using %3Q and %3N to no avail, both caused a ms value of 4 to be 400 and 91 to be 910 )

In reading other comments on here, i think i need to use a custom datetime.xml file to handle this... so i wrote customdatetime.xml referenced above in my props.conf and detailed here:

<datetime>
    <define name="test_dateformat" extract="year, month, day">
        <text><![CDATA[(\d{4})-(\d{2})-(\d{2})]]></text>
    </define>
    <define name="test_timeformat" extract="hour, minute, second, subsecond, zone">
        <text><![CDATA[(\d{2}):(\d{2}):(\d{2}):(\d+)(-\d{2}:\d{2})]]></text>
    </define>
    <timePatterns>
        <use name="test_timeformat" />
    </timePatterns>
    <datePatterns>
        <use name="test_dateformat" />
    </datePatterns>
</datetime>

That does not seem to work either. Where am i going wrong?

Re: Non-Padded Milliseconds on Timestamp

ivanreis — Wed, 30 Sep 2020 02:34:38 GMT

Hi darrenfuller, I was reviewing the splunk docs and I would like to suggest the configuration below:

TIME_FORMAT = %Y-%m-%dT%H:%M:%S:%3Q%:z
Where : %:z is to specify hour and minute separated by a colon

TIME_FORMAT = %Y-%m-%dT%H:%M:%S:%Q%:z

%z The timezone offset from UTC, in hour and minute: +hhmm or -hhmm. For example, for 5 hours before UTC the values is -0500 which is US Eastern Standard Time.
Examples:

Use %z to specify hour and minute, for example -0500
Use %:z to specify hour and minute separated by a colon, for example -5:00
Use %::z to specify hour minute and second separated with colons, for example -05:00:00
Use %:::z to specify hour only, for example -05

Also comment #MAX_TIMESTAMP_LOOKAHEAD = 40, because maybe splunk is looking for the date/time on a position where this data is not found, because generally the date/time is being on the begin of the log. This is only a suggestion for troubleshooting. If it works, you can try to play with numbers to identify the right characters positions to get this data properly indexed and re-enable it again.

MAX_TIMESTAMP_LOOKAHEAD = Specify how far (how many characters) into an event Splunk software should look for a timestamp.
This constraint is applied starting from the location positioned by TIME_PREFIX.

For example, if TIME_PREFIX positions a location 11 characters into the event, and MAX_TIMESTAMP_LOOKAHEAD is set to 10, timestamp extraction will be constrained to characters 11 through 20.

If set to 0 or -1, the length constraint for timestamp recognition is effectively disabled. This can have negative performance implications which scale with the length of input lines (or with event size when LINE_BREAKER is redefined for event splitting).

I hope this can assist you to fix your issue.

Re: Non-Padded Milliseconds on Timestamp

darrenfuller — Mon, 21 Oct 2019 16:25:28 GMT

Thanks for your comment. I tried both of these configs and neither worked. The result was the same for both of the examples above with < 3 digits on the millisecond...

"2019-10-08T14:47:33:4-04:00" was transformed into "2019-10-08T14:47:33:400-04:00"

and

"2019-10-08T14:47:29:91-04:00" was transformed into ""2019-10-08T14:47:29:910-04:00"

Re: Non-Padded Milliseconds on Timestamp

darrenfuller — Mon, 21 Oct 2019 16:27:38 GMT

and my TIME_PREFIX is time=" which only occurs towards the end of these events . if you start there... 40 characters is more than enough to match these timestamps (which are 27-29 characters in length)

Re: Non-Padded Milliseconds on Timestamp

darrenfuller — Fri, 22 Nov 2019 17:46:30 GMT

I have decided (due to the lack of answers) that this is in fact not possible.