How to Interpret the Munin Memory Graphs

I just recently installed Munin on my Ubuntu 10.04 VPS to help me visualize resource usage.

A lot of the graphs it creates are fairly straightforward, but I'm afraid I've met my match with the memory usage ones. Does anybody know how to read these? And how to interpret the graphs below?

~~![](<URL url=)http://img84.imageshack.us/img84/6149/memoryday.png" />

![](" />~~

3 Replies

I think you may need to offer more information on just what you are having trouble reading (do you have a question on a specific stat?), but in general, the first set of information, apps through swap, are stacked on the graph as color regions. (Some may be hard to see if very small). All but swap summed together has to fit in physical memory, which is why your overall graph height is at 512MB.

For example, on your system, all of your apps are only using about 100MB, with the rest of your memory largely as disk page cache or filesystem buffers, which is Linux taking advantage of your otherwise unused memory. The sawtooth pattern you see on the week graph, where unused grows suddenly (and cache decreases) could be a periodic process restart or system activity that flushes the cache.

The next set of information - inactive through mapped - are shown as lines on the graph superimposed on the previous regions, on the same scale. They aren't additive (stacked).

If you're looking for detail on what the individual stats mean, you should have a key at the bottom of the munin graph page. Also, these are stats pretty much taken directly from /proc/meminfo, so any Linux documentation on that can be used as well.

– David

Thanks David! Your explanation really helped. It also confirmed what I thought, but was unsure of – that apps are only using 100mb of memory. I guess I won't be needing a 768 anytime soon :)

Yeah, probably not yet with those stats.

If you just want a quick instantaneous answer along those lines, another option is to use "free -m" and look for the used column without buffers/cache. For example, on one of my Linode 512 nodes (not yet rebooted since the recent plan upgrades so still essentially a Linode 360):

~$ free -m
             total       used       free     shared    buffers     cached
Mem:           360        345         15          0        189        108
-/+ buffers/cache:         46        313
Swap:          255          8        247

So on this node (a static web site) applications need about 50MB of memory with the remainder going towards buffers or cache. You should find that the output on your system matches your ~100MB from munin fairly well at least in steady state.

Of course, that's a snapshot, so something like munin lets you watch it over time.

– David

Reply

Please enter an answer
Tips:

You can mention users to notify them: @username

You can use Markdown to format your question. For more examples see the Markdown Cheatsheet.

> I’m a blockquote.

I’m a blockquote.

[I'm a link] (https://www.google.com)

I'm a link

**I am bold** I am bold

*I am italicized* I am italicized

Community Code of Conduct