Need help using thold to monitor used ram

[shadv]

I am actively monitoring RAM from Windows servers using THold/Monitor on Linux. I'm just using the standard graphs and then created a Threshold template based on a percentage free.

Hello NetniV,

Thank you for the reply, I wish it was that easy on linux, however when you have a lot of read/writes in linux your Cache can take up all of the Free ram, this is still treated as Free ram by the linux system. I would monitor the Cache, however this can change depending on the amount of read/writes at the time, so the only way I can think of is to either monitor the Cache+Free or the Used on the graph I use, Thold will not monitor the Used memory being it is not a data source, and I can not figure out a way to add the data sources together to monitor them.

[netniV]

I think the net-snmp templates do break out the linux memory usage properly (at least I thought so). I'd have to look at one I setup as a test a while ago and see if it's still configured. If it is, that may help you.

[shadv]

I guess there is not any way to do this reliably on a Linux server then?

[netniV]

I think you miss understood what I was meaning there. I run Cacti ONLY on Linux, but I grab memory from Windows servers and Linux Servers. The NetSNMP templates are different to the standard hmib-based templates in they create a stacked graph like what you are after.

[shadv]

Thank you for the reply,

Do you have a graph that works for this? The build in graph does not work, nor do any that I have tried so far.

The ones I have tried break it down to memory free, cache, total, and buffer. None of these will work for reliable threshold monitoring in my systems.