[mrtg] Re: migration from NT to Linux

[Scheidel, Greg]

> The problem is different.

Maybe so, maybe not.  Since you do not directly address my point, its hard
to say.

> You answer something not related to the question.

I don't believe so.  He asked for general information regarding moving log
files from NT to Linux; I answered with what I believed was a relevant
concern.  Perhaps you should reread his original post (handily enough, you
included it in your reply).

> Also, if you are referring to the MRTG docs, reread them.

I have, and I stand by my statements.  If you have something *specific* to
add or point out, rather than vague generalizations intimating that I have
not read the docs, please do so.

> I quote:

> > Hi,
>>
>> Is there any difficulties to migrate MRTG from NT to a Linux ?
>> I want to keep all my *.log files
>>
>> Regards,
>>
>> Oliver somewhere in The Hague
>
>No relation whatsoever to that formula.

It is, in that he asked for general information regarding moving the log
files from NT to Linux, and that I had found the formula in the MRTG docs to
not match that on my Linux box.

> And I quote again:

Actually, this is the first time you quoted.  Before you just referred.

>>          A.  The Unix timestamp (1.A) when the data on
>>              this line was gathered.
>>
>>              This timestamp may be converted in EXCEL by
>>              using the following formula and choosing the
>>              option for the "1904 base":
>>
>>                     =(x+((365*66+17)*86400))/86400
>>
>>              If the "1900 base" is selected, then change
>>              the "66" value to "70".
>
>*I* did *not* insert EXCEL here.

You *did* insert EXCEL in your response to me when you said (referring to
the formula for converting from non-skip seconds to readable time format):

> First of all, the logfile should be the same as far as I know.  Secondly,
above formula is for excel, not for unix timestamps.

It was my point, in response to that, that using Excel was largely
irrelevant.

> There are two problems with this formula:
>
> 1) It converts GMT times to GMT times.  You are somewhere in the U.S.
>   and hence the 5/24 you use in your formula.  This is only valid when
>   daylight saving is not (or: is) applied and only if you want to see
>   your timezone.  For me, the formula as you provide is not right.
>
> 2) The formula adds a fixed number of leap days to the 66 or 70 years.
>   This number is not correct anymore; it should be 19 nowadays, indeed.
>
>   Perhaps the following formula will do:
>   =(x+DATE(1970,1,1)-5/24) for timezones 5 hours west of Greenwich
>   =(x+DATE(1970,1,1)+1/24) for Europe
>   and so on.

Ah, something relevant and specific to add to the conversation.  Let me see
if I understand your points:

1) The timestamp as listed in the MRTG log files is GMT.  So when I convert
it to a readable time format on my Linux box, it automatically adjusts for
GMT to my time zone conversion (since my Linux box is after all correctly
configured for my time zone) so that I don't realize the difference.  But
when I do the conversion by hand (or in Excel) I come up with a readable
time that does not match what I think is correct, as I have overlooked the
time zone difference.  Thus I adjust the time thinking that it is a possible
issue with the Linux box, when in reality it is my error in not taking into
account the time zone difference.

2) The fixed number of days to correct for leap year is in fact now 19 (and
soon will be 20).

So basically I screwed up by not taking into account GMT but adjusted for it
without realizing what it was I was adjusting for. :)

Greg S. 

--
Unsubscribe mailto:mrtg-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:mrtg-request at list.ee.ethz.ch?subject=help
Archive     http://www.ee.ethz.ch/~slist/mrtg