Hi Ruben,

Yes, I definitely agree :-) Thinking at it, it should be good idea to
include timezone information there. Noting this one down.

Cheers,
Paolo

On Wed, Apr 11, 2012 at 01:01:50PM +0200, Ruben Laban wrote:

> It's actually much simpler. Each file already contains the required info  
> (though its presence does depend on your config). For example:
>
> --START (1334141100+300)--
>
> Start + interval = all the (timestamp related) info I need
>
> No need to check file ctimes and/or parse configs. I'm guessing relying  
> on this piece of information is as good as any other solution, if not  
> better.
>
> Regards,
> Ruben
>
>
> On 29/Mar/2012 23:02, Paolo Lucente wrote:
>> Hi Ruben,
>>
>> The risk of #2 is that if anybody accidentally does anything with
>> the file, you might be screwed. It's also true you insert data in
>> RRD shortly after the file is created so the probability is slim.
>> But in this sense you can perhaps consider relying on file ctime
>> rather than mtime.
>>
>> Relying on the file name is IHMO the way to go. Depending on how
>> you are inserting data in a RRD file, your option #1 looks OK to
>> me (granted that you define somewhere the config in use and parse
>> it accordingly - or more simply you define the step explicitely).
>>
>> Cheers,
>> Paolo
>>
>> On Thu, Mar 29, 2012 at 12:42:05PM +0200, Ruben Laban wrote:
>>>
>>> Hi Paolo,
>>>
>>> Thanks for the clarification.
>>>
>>> Now to think of a nice way to deal with this "issue".
>>>
>>>   * Manually add the interval to the file's timestamp, or
>>>   * Rely on the file's mtime, or
>>>   * Ignore the problem and insert "inaccurate" data into the RRDs
>>>
>>> None of those are really idea, tho I guess the 2nd one will turn out to
>>> the most reliable.
>>>
>>> Regards,
>>> Ruben Laban
>>>
>>> On 29/Mar/2012 09:01, Paolo Lucente wrote:
>>>> Hi Ruben,
>>>>
>>>> That is intended. What is encoded in the filename is the basetime of a
>>>> timeslot, what in a SQL table would be the stamp_inserted field. The OS
>>>> timestamp of the file, ie. when the file was last touched, encodes what
>>>> in a SQL table would be the stamp_updated field instead.
>>>>
>>>> Cheers,
>>>> Paolo
>>>>
>>>> On Wed, Mar 28, 2012 at 11:14:48AM +0200, Ruben Laban wrote:
>>>>> Hi Paolo, list,
>>>>>
>>>>> Just now I noticed another unexpected behavior:
>>>>>
>>>>> -rw------- 1 root root 402 2012-03-28 10:50
>>>>> pmacct_up0_in_total_20120328-1045
>>>>> -rw------- 1 root root 402 2012-03-28 10:55
>>>>> pmacct_up0_in_total_20120328-1050
>>>>> -rw------- 1 root root 402 2012-03-28 11:00
>>>>> pmacct_up0_in_total_20120328-1055
>>>>>
>>>>> The documentation states:
>>>>>
>>>>> KEY:            [ sql_table | print_output_file ]
>>>>> DESC:           ... Dynamic names are supported through the use of
>>>>> variables, which are computed at the moment when data is purged to the
>>>>> backend. ...
>>>>>
>>>>> So I'd expect the file written at 10:50 to called -1050, not -1045.
>>>>>
>>>>> Am I interpreting the documentation wrong, did I stumble upon a bug, or
>>>>> something completely different?
>>>>>
>>>>> Regards,
>>>>> Ruben Laban
>>>>>
>>>>> _______________________________________________
>>>>> pmacct-discussion mailing list
>>>>> http://www.pmacct.net/#mailinglists
>>>>
>>>> _______________________________________________
>>>> pmacct-discussion mailing list
>>>> http://www.pmacct.net/#mailinglists
>>>
>>
>> _______________________________________________
>> pmacct-discussion mailing list
>> http://www.pmacct.net/#mailinglists
>

_______________________________________________
pmacct-discussion mailing list
http://www.pmacct.net/#mailinglists

Reply via email to