System monitoring tools - Re: [SGVLUG] unstable ubuntu system

[Claude Felizardo]

On Fri, Feb 22, 2008 at 3:16 PM, matti <mathew_2000 at yahoo.com> wrote:
>
>  Hi!
>
>  OK!
>
>  So, we suspect the instability to be due to limited
>  RAM resources.... SO how do we confirm this?
>
>  I recall using some tools back in my SGI super
>  computing days which were really really nice,
>  as the system could generate a log file and
>  you could see what the bottle neck was...
>
>  I thought SGI was going to open source it..
>  and looking at
>
>  http://oss.sgi.com/projects/
>
>  it looks like they may have something
>  (not the exact item I used back in "the day")
>  http://oss.sgi.com/projects/pcp/
>
>  Question -
>
>  What monitoring tools do people here like to use?
>
>  personally, the standard I've seen include with
>  Ubuntu leave more to be desired.
>  (I would like to see a history log of the graphs)
>
>  thanks!
>  matti
>
>  ps - I've managed to find an additional 128MB RAM
>  to put into the system (upto 384MB now.. still short
>  of 512MB, but I only have 2 sdram slots.. )
>

so you suspect you might not have enough memory so now you want to run
some monitoring software on it?   ;)

i've been using systemgraph from http://www.decagon.de/sw/systemgraph/
for a while now.  I gave a demo of it not that long ago.  you can
selectively enable graphing of ram/swap, cpu info, hd temp and io, num
open files, load avg, networking stats, etc.  Check out the guy's
website, it has lots of pretty pics.  By default it will show last 4
hours but it can show anywhere from 1 hr to 365 days - oh, you can ask
it to show the last 45 seconds if you want though it normally samples
stuff at 1 and 5 minutes intervals depending on what it's monitoring.

As for reqs, it uses RRDTool, perl, iostat, lm_sensors, hddtemp,
sysstat, httpd, etc.

oh hey, looking at the plots from my soon-to-be replaced file server
(p3-450 w/ 64MB), I can see that the used swap would start to climb
just before it would either lock up or the kernel would start to kill
tasks randomly.  Unfortunately it's my two media server apps that are
the memory hogs - one java based, the other perl based.

claude