Munin :: 95th Percentile Calculations
Dec 5, 2008I have Munin running on my Centos 5.2 box, but would like the 95th percentile calculations. I have found this page for an add on:
[url]
where to put this code and/or how to install it?
I have Munin running on my Centos 5.2 box, but would like the 95th percentile calculations. I have found this page for an add on:
[url]
where to put this code and/or how to install it?
I have a question about 95th percentile billing. Say I have a 6mbit commit. That means I should be getting 2000GB/month. I calculated that if I burst 100mbit until I get 2000GB then do no bandwidth for the rest of the month, I would get charged for 100mbit of commit, is this correct?
Calculations
2000000MB * 8 = 16000000 mbits / 100mbit = 160000 seconds / 60 = 2667 mins / 60 = 44.4
So 44.4 hours of 100mbit equals 2000GB.
Now lets say they take a snapshot every 5 mins.
44.4 hours * 60mins = 2666 / 5 mins per snapshot = 533 snapshots
Now lets find the total number of snapshots in a month.
30 days * 24 hours * 60 minutes / 5 minutes per snapshot = 8640 snapshots
Now we find 95th percentile
8640 * .95 = 8208
8640 - 8208 = 432
So 432 snapshots will be dismissed as burst, but I will have 533 snapshots at 100mbit (as calculated above).
95th percentile...
what is it?
Is it possible to have a 40Mbps 95th percentile, but only push 3TB in a month?
I know 40Mbps=~12,000GB, that's why it doesn't make sense.
Basically provider offers 2000gb bandwidth.
However, at the end of the month they don't look at the total X GB's transfered, they measure it up with 95th percentile.
I know there's a formula involved, but maybe someone can give me some insight on why 95th is for, and the formula of course would help
100mbps can be rather vague at first so I will now try breaking 100mbps, - my question will be near the bottom.
Ive used a conversion calculator to draw this up.
8192 Mbit (Megabits) = 1 GB (GigaByte)
100 Mb (Megabits) = 0.01220703125 (GigaBytes)
So, to reach 1 GB it will take just under 1 min 22 seconds at a rate of 100mbps.
8192 Mbit / 100 Mbit = 81.92 seconds which is 1 min 21.92 seconds.
1 min 21.92 seconds = 1 GB.
My question:
As i am more familiar dealing with Gb's more so than Mbits i have a few questions which i am not so sure about .....
On the 95th percentile model, I recall reading bursting not calculated for 36 hrs in this model.
In our application, for now, we will only be uploading to the colocation 6 days a week, on a 10 meg upstream - averaging anywhere from 2 hrs nightly to Saturday ....real small upload.
When I look at 95th percentile, I am looking at it that we would be billed at 9Mbps if we were to utilize this billing method as opposed to purchased transit.
some light on the math & or the *36 hr* burstable bandwidth ?
Just curious about what kind of pricing people have been able to get when going for 95th percentile billing, and what kind of quantities?
For example, currently we are paying $44/Mbps for a quantity of 250Mbps, so $11k/month for roughly 100TB of transfer.
Does that seem like a really high price?
Why do seemingly most companies do 95th percentile billing based on a 5 minute polling interval?
Do you think there would be any complaints about basing 95th percentile billing on a 1 minute polling interval?
I just had a quick question, I have been using dedicated servers for a long time now, and I was just looking to host one of my own 1U servers that I have been using at home, now I had a simple question, but it seems to confuse me on what exactly do I pay at the end of the month.
I have read about the 95th percentile billing, and paying for 1Mbps.
I usually had 1000GB with my servers, simple, easy.
All I need to know is if I get 1Mbps, can I not go over that 1Mbps while backing up files, or hosting a game server? or is it that I have a certain amount of bandwidth because this is really confusing me.
1. I have loads of these showing up in the WHM Apache service status
52-1 - 0/0/8 . 0.00 3970 0 0.0 0.00 20.02 127.0.0.1 server.mybox.com OPTIONS * HTTP/1.0
53-1 - 0/0/6 . 0.00 3969 0 0.0 0.00 79.77 127.0.0.1 server.mybox.com OPTIONS * HTTP/1.0
54-1 - 0/0/5 . 0.00 3968 0 0.0 0.00 38.04 127.0.0.1 server.mybox.com OPTIONS * HTTP/1.0
I was told that was normal as its apache's MPM prefork way of waking up the worker processes. That I have accepted however it has been problematic ever since I installed mod_security that things have been going ape.
2. The Bandwidth in CPanel and that in Apache does not add up. There is only 1 account on the server. The bandwidth usage in CPanel shows 51GB for the month of July so far, however in WHM under Apache status it shows 223GB of Bandwidth in the last 48 hours.
Now given the the host tells me that CPanel does not count certain bandwidth(?) I am not certain what to think as the difference in bandwidth is a whopping 160GB in the past 2 days.
WHM
Server uptime: 1 day 17 hours 54 minutes 31 seconds
Total accesses: 87589 - Total Traffic: 222.5 GB
CPU Usage: u110.4 s13.25 cu0 cs0 - .082% CPU load
.581 requests/sec - 1.5 MB/second - 2.6 MB/request
15 requests currently being processed, 3 idle workers
CPanel
Monthly Bandwidth Transfer 51915.52
Total50.7 Gig
Server Load 1.7GHZ Machine with 1GB Ram
Server Load 0.25 (1 cpu)
Why such a HUGE difference in bandwidth?
How do you deal with off-site backups on a colo machine billed at 2 Mbps 95 percentile?
Moreover, do you do any httpd throttling so a few spikes don't really hurt you on the 95% billing method?
I have WHM 11.15.0 cPanel 11.17.0-R19429
FEDORA 4 i686 on standard - WHM X v3.1.0
/var keeps getting filled up with munin logs. Do I really need these logs? The server already had munin when I got it so I don't know how it was installed.
Can I disable it? I looked at the cron tab, and I don't see munin, but when I did chkconfig --list I found:
munin-node 0: off 1: off 2: on 3: on 4: on 5: on 6: off (spaces added because it made smileys)
How do I disable it?
How to disable from debian server autorun munin script [url] [munin, munin-update, munin-node processes]?
View 13 Replies View Relatedif you can dump the actual data used to make the graphs in munin, or are the graphs the only way to view the data?
View 2 Replies View RelatedI tried to install cacti, somehow it didn't work out so i decided to install munin. But when i choose munin from plugins list it starts to install than it says
Starting Munin Node: [ OK ]
Can't locate RRDs.pm in @INC (@INC contains: /usr/lib/perl5/site_perl/5.8.8/i386-linux-thread-multi /usr/lib/perl5/site_perl/5.8.7/i386-linux-thread-multi /usr/lib/perl5/site_perl/5.8.6/i386-linux-thread-multi /usr/lib/perl5/site_perl/5.8.5/i386-linux-thread-multi /usr/lib/perl5/site_perl/5.8.8 /usr/lib/perl5/site_perl/5.8.7 /usr/lib/perl5/site_perl/5.8.6 /usr/lib/perl5/site_perl/5.8.5 /usr/lib/perl5/site_perl /usr/lib/perl5/vendor_perl/5.8.8/i386-linux-thread-multi /usr/lib/perl5/vendor_perl/5.8.7/i386-linux-thread-multi /usr/lib/perl5/vendor_perl/5.8.6/i386-linux-thread-multi /usr/lib/perl5/vendor_perl/5.8.5/i386-linux-thread-multi /usr/lib/perl5/vendor_perl/5.8.8 /usr/lib/perl5/vendor_perl/5.8.7 /usr/lib/perl5/vendor_perl/5.8.6 /usr/lib/perl5/vendor_perl/5.8.5 /usr/lib/perl5/vendor_perl /usr/lib/perl5/5.8.8/i386-linux-thread-multi /usr/lib/perl5/5.8.8 . /usr/local/cpanel/3rdparty/lib/perl) at /usr/share/munin/munin-graph line 225.
BEGIN failed--compilation aborted at /usr/share/munin/munin-graph line 225.
Install Complete (munin 1.2.4)
Done
I did some googling about the error and it is because rrdtool is not installed.
Actually i installed rrdtool 1.3.0 from its website it was in /usr/local/rrdtool
Anyway i clicked install a rpm from whm and choose rrdtool.i386
root@server # rpm -qa | grep rrd
rrdtool-1.2.23-6.fc6
But it is still giving error above when i try to install munin
Munin was running OK on my cPanel server until yesterday when the graphs stopped updating themselves. I attempted a "/usr/share/munin/munin-update --force-root" which did nothing. Then I uninstalled and reinstalled munin on the cPanel plugins screen and now its running but all the graphs are empty
View 3 Replies View RelatedI use Munin to monitor the health of our servers, I can tell by looking at the graphs there's nothing to worry about, however, I'm struggling to baseline acceptable performance.what would be classed as 'normal' output for some of the more relevant munin graphs.
I've been looking at the Apache* modules and this is the output from one of our servers:
average of: 300 accesses per minute, 6 busy servers and 4.10MB a minute volume
max of: 1400 accesses per minute, 81 busy servers and 51MB a minute volume
This is a dedicated box running one site.
We have another box that is running approximately 30 sites
average of: 30 accesses per minute, 1 busy server and a 500K a minute volume
max of: 322 accesses per minute, 11 busy servers and a 4MB a minute volume.
These servers are pretty much the same spec, dual core 64Bit, 4GB of ram, two SATA disks in RAID1.
I'm checking my CPU/Memory/MySQL Usage and I'm seeing that something called Munin is running at 54%. That is apparently the average for today. Can anyone give me any insight as to what this is?
Wikipedia classifies Munin as a network monitoring application. I don't think it should be using this much processing.
It appears that I may have enabled this plugin when I rebuilt my apache using EasyApache. Are there any big benefits to this? Looking at the graphics at first glance it doesn't look like there is. but then again, I'm not that knowledgable about this stuff.
Does anyone have a tutorial or instructions to show how to set up munin to monitor bandwidth/traffic utilization?
I have a VPS from a provider that has no monitoring tools, and I need something to monitor.
(The VPS uses the HyperVM software.)
I am using the CentOS version 4.x right now, and munin is already set up and configured with a basic install.
Anyone know how to add traffic/bandwidth monitoring? (Ideally, it would monitor all active ports.. but I only really need it to monitor port 1194 for openVPN software.)
I rebooted my server and now munin is not showing any eth0 traffic. All other graphs are fine. I can see there is a ton of apache accesses so there is definitely eth0 traffic. Munin logs report no errors. I restarted munin and munin-node. I even did 'yum remove munin munin-node' and reinstalled again but it still doesn't work for eth0.
View 4 Replies View RelatedThe memory usage on my server has slowly been creeping up over the last few weeks due to increased load. Here is my munin memory usage graph for the last week:
[url]
Could someone tell me what I need to be looking for or how to interpret this?
The cache and buffers have been getting really high but also has the inactive memory… so its looks as though its using memory and marking it as inactive afterwards and although graph shows I have massive memory use at first glace it also shows I have half my ram inactive and therefore available to be used.
On Cpanel/WHM. I have just moved from a VPS to a dedicated server. I reinstalled munin, so get some stats via that. I used to have apachetop loaded on my VPS for when I wanted a 'near realtime' streaming view of apache access.
I'm wondering what the best solution is to get a good view of apache, like what apachetop did, plus also it would be nice to have a real-time monitor of MySQL activity, HDD activity (such as I/O queues, etc. Something along the lines of the perfmon on Windows servers.
What is my best option?
Also, with Nagios, when I look at the website, it seems there are two options. Load it on a single server and then load the stats via [url]or have the Nagios 'stat collector' on one machine, and have it gathering stats from multiple machines.
If you only install it on a single dedicated server, do you really have to be on the console and connect to the Nagios stats via localhost, rather than connecting remotely?
Ideally, I would like a quick, easy to setup solution, but if it takes some configuring, I can deal with it, as long as there is some documentation. My main goal is to get the real time type monitoring, you get with window's perfmon.