[syslog-ng] Archive module ?

[Scot Needy]

Thanks Evan I will try your advice and log directive. 



> On Sep 8, 2016, at 10:31 PM, Evan Rempel <erempel at uvic.ca> wrote:
> 
> Your original question was "Is there a better way to archive my syslog data", but attempting to answer your question anyway.
> 
> If the source is not syslog data, but you are using syslog-ng to read the data and place it into some storage platform (splunk, database or elasticsearch), then by definition it is syslog data. There will be default program, facility, level on each event. Writing this data to a file with my suggested format will allow for replaying of the log data. For the non-syslog sources, the replay of the syslog stream is functionally equivalent to re-reading the source data. It is just an intermediate format of the data.
> 
> This is not limited to syslog data. It can be applied to any time series event style data. This covers all logging data and also audit data (audit "log" really is "log" data).
> 
> The challenge with using blocks of raw data is that you require the SAME storage engine when replaying your logs. Even different versions of the storage engine may have different raw formats. using a text file for archiving would allow replaying the logs even years later when a different storage and presentation framework is used. We have had to reply logs as far back as 7 years for audit reporting.
> 
> I hope this provides some deeper rationale for the text file format.
> 
> Evan.
> 
> On 09/08/2016 06:56 PM, Scot Needy wrote:
>> 
>> That looks like it would work but if the source isn’t syslog data ? 
>> 
>> I pondered this question when looking at the disk buffer settings in 3.8. 
>> If my data is in ES, MySQL or some other parsed presentation platform then all I really need to write out blocks of raw data to disk and archive for SOX compliance. 
>> 
>> Let the storage archive, compress, encrypt and de-dup. 
>> 
>> Just 
>> 
>> 
>>> On Sep 8, 2016, at 5:48 PM, Evan Rempel < <mailto:erempel at uvic.ca>erempel at uvic.ca <mailto:erempel at uvic.ca>> wrote:
>>> 
>>> You would have to use a syslog-ng template defined as
>>> 
>>> "<$PRI>$ISO_DATE $FULLHOST $MSGHDR$MESSAGE\n"
>>> 
>>> and that would be the "on the wire" format. The you can netcat this to the syslog port which would replay the log.
>>> 
>>> Evan.
>>> 
>>> On 09/08/2016 02:39 PM, Scot Needy wrote:
>>>> Thanks Even, 
>>>> 
>>>>  Maybe I’m doing something wrong.   
>>>> I see some of the headers but not all. 
>>>> 
>>>> 2016-09-08T17:31:17-04:00 <IPADDR> <HOST> %ASA-3-305006:  regular translation creation failed for imp src …. 
>>>> 
>>>> I guess what I was looking for is something that I could write blocks of data to and read from irregardless of the type of data. 
>>>> Something almost like tcpdump or buffer that could rotate every hour or so and then be used to re-play directly back into syslog-ng. 
>>>> 
>>>> Log > YYYY.MM.DD.HH_compressed_data. 
>>>> 
>>>>   
>>>> 
>>>> 
>>>> 
>>>>> On Sep 8, 2016, at 5:13 PM, Evan Rempel <erempel at uvic.ca> <mailto:erempel at uvic.ca> wrote:
>>>>> 
>>>>> If you write the syslog data to a file with all of the data, then it can be ingested again.
>>>>> 
>>>>> We save the data to the files like
>>>>> 
>>>>> 2016-09-08T00:00:01-07:00 host.name.here mail.info <http://mail.info/> sm-mta[9521]: u887019I009521: milter=mimedefang, action=rcpt, continue
>>>>> 
>>>>> and then we have a syslog-ng pattern database that will turn this back into an "on the wiire" format that can be sent to any syslog port.
>>>>> 
>>>>> Alternatively, you could save it in the file as the original format, and then it can just be sent back to syslog when you need to. Human readablility is not quite as good, but if you have the data in splunk or whatever, then humans don't need to read the raw files.
>>>>> 
>>>>> Evan.
>>>>> 
>>>>> 
>>>>> On 09/08/2016 01:32 PM, Scot Needy wrote:
>>>>>> Hi All,
>>>>>> 
>>>>>> Is there a better way to archive my syslog data once it’s been received and injected into elk,spunk or whatever ?
>>>>>> 
>>>>>>  I see how I can log my data to flat files using date macros then compress and archive as normal data but is there a better way to write out the data in the same format it was received to facilitate re-injesting ?
>>>>>> 
>>>>>> For example: My syslog data on file does not include the original syslog header data .
>>>>>> 
>>>>>> Syslog is just one example.
>>>>>> 
>>>>>> 
> 
> ______________________________________________________________________________
> Member info: https://lists.balabit.hu/mailman/listinfo/syslog-ng
> Documentation: http://www.balabit.com/support/documentation/?product=syslog-ng
> FAQ: http://www.balabit.com/wiki/syslog-ng-faq
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.balabit.hu/pipermail/syslog-ng/attachments/20160912/26238c50/attachment.htm