I just installed MRTG yesterday with the help I found here (http://webhostingtalk.com/showthread.php?threadid=20548) (thanks RutRow)... Now some questions:

1. Why is it not advisable to run mrtg as root?

2. My MRTG graph kinda looks strange to me (www.dnspropagation.net/mrtg/), since the incoming traffic into the server is about 1/3 that of outgoing traffic. FTP/email transfer do not seem to explain this, especially after looking at the mrtg graph for the sites hosted on the server (with practically zero incoming traffic). Hence, what is the source of this incoming traffic?

3. The MRTG (average monthly) does not seem to match the report made by webalizer. I know that bandwith measured by average method will be the same (or about the same) as the actual data transfer. Even if I cosnider outgoing traffic only (monthly average), it's about 50GB of transfer/mo yet webalzier claims only around 35GB ... Can anybody explain the difference?
BTW, MRTG and bandmin now report about the same number! With respect to this, I would like to see about any reports where MRTG and bandmin differ. Both MRTG (with the patched version given by RutRow, see link above) and bandmin record bandwith the same way (getting info from eth card, I believe), but webalizer only analyze data on http log. So why is webalizer to be trusted moren than bandmin or MRTG?

That's all for now ... if I have further question, I will post here. Thanks in advance

cheers,
:beer:

Originally posted by dektong
1. Why is it not advisable to run mrtg as root?

1. You should never run anything as root unless it is absolutely necessary.
2. It is not absolutely necessary.


2. My MRTG graph kinda looks strange to me (www.dnspropagation.net/mrtg/), since the incoming traffic into the server is about 1/3 that of outgoing traffic. FTP/email transfer do not seem to explain this, especially after looking at the mrtg graph for the sites hosted on the server (with practically zero incoming traffic). Hence, what is the source of this incoming traffic?


Hard to say without looking at graphs for the separate traffic types, but judging by your URL I'd guess maybe lots of DNS traffic?


3. The MRTG (average monthly) does not seem to match the report made by webalizer. I know that bandwith measured by average method will be the same (or about the same) as the actual data transfer. Even if I cosnider outgoing traffic only (monthly average), it's about 50GB of transfer/mo yet webalzier claims only around 35GB ... Can anybody explain the difference?


MRTG and webalizer are measuring different things. Imagine sending a 10 lbs glass sculpture through the mail; this is quite fragile, so you want to add lots of packing material to keep it safe. In the end you have a 20 lbs package containing the sculpture plus 10 lbs of packing foam.

MRTG, because it looks at the outside of packages, will count this as 20 lbs. Webalizer, because it is completely unaware of how you are packaging the HTTP data, counts this as 10 lbs.

Cperciva, thanks for replying ...

Originally posted by cperciva


1. You should never run anything as root unless it is absolutely necessary.
2. It is not absolutely necessary.

ah ... I see ... OK, will change ... not sure how, but will find out.



Hard to say without looking at graphs for the separate traffic types, but judging by your URL I'd guess maybe lots of DNS traffic?


If it is OK with you, I will contact you in private since I am not sure whether it's good to give my client's mrtg without their permission ... About the DNS traffic, I am not sure about that, since all the sites hosted on the server do not use the DNS in that server, but on my other server ...


MRTG, because it looks at the outside of packages, will count this as 20 lbs. Webalizer, because it is completely unaware of how you are packaging the HTTP data, counts this as 10 lbs.


This I totally agree ... except some people said that bandmin is not trustworthy and even claim their MRTG and webalizer report the same thing, and this is where I got confused ... So far my bandmin and MRTG report very close numbers (negecting my incoming traffic, just outgoing traffic) yet bandmin/MRTG and webalizer differ by a factor of almot 3 (webalizer is less).

Again, thanks a lot

cheers,
:beer: