System Resources Threshold Values ???

Config: HP 9000 Server N Class Server
HDD-2*9Gb Ultra 2 SCSI Internal HDD
RAM-1GBMemory

Can anybody tell me what are the Parameters should be Monitored and their Threshold Values.
I`m curently montoring these values thru Sar,vmstat,top.But I`m not aware how much it should be .
OR Guide me where I`ll find these Values, so that if these values crosses the certain extent,Action has to be taken.

For Memory , CPU , DISK , CACHE ,Network
Help will be Appreciated.
Thanx
Rgds
Jaikishan

Don`t Give-Up

7 REPLIES 7

Hi
IMHO

It is very difficult to say what a threshold value should be as it varies for each server and its environment.

Each sysadmin gets a "Feel" for their machines.

It is normal to say that above 4.00 on top and the server is loaded - earlier this year I had to run a server that was averaging 12.00 - yes it was slower but we were still able to do business, it did runaway once and hit 154.60 before I could bring it under control (Lots of fun).

I do monitor Memory , CPU , DISK , CACHE and Network activity, but could not say that at ?X? value for a particular area I have a problem.

I tend to log in as a normal user and am aware of the ?Feel? of my machines.

Paula

If you can spell SysAdmin then you is one - anon

Hi
IMHO

It is very difficult to say what a threshold value should be as it varies for each server and its environment.

Each sysadmin gets a "Feel" for their machines.

It is normal to say that above 4.00 on top and the server is loaded - earlier this year I had to run a server that was averaging 12.00 - yes it was slower but we were still able to do business, it did runaway once and hit 154.60 before I could bring it under control (Lots of fun).

I do monitor Memory , CPU , DISK , CACHE and Network activity, but could not say that at ?X? value for a particular area I have a problem.

I tend to log in as a normal user and am aware of the ?Feel? of my machines.

Paula

If you can spell SysAdmin then you is one - anon

Hi
IMHO

It is very difficult to say what a threshold value should be as it varies for each server and its environment.

Each sysadmin gets a "Feel" for their machines.

It is normal to say that above 4.00 on top and the server is loaded - earlier this year I had to run a server that was averaging 12.00 - yes it was slower but we were still able to do business, it did runaway once and hit 154.60 before I could bring it under control (Lots of fun).

I do monitor Memory , CPU , DISK , CACHE and Network activity, but could not say that at ?X? value for a particular area I have a problem.

I tend to log in as a normal user and am aware of the ?Feel? of my machines.

Paula

If you can spell SysAdmin then you is one - anon

Hi
IMHO

It is very difficult to say what a threshold value should be as it varies for each server and its environment.

Each sysadmin gets a "Feel" for their machines.

It is normal to say that above 4.00 on top and the server is loaded - earlier this year I had to run a server that was averaging 12.00 - yes it was slower but we were still able to do business, it did runaway once and hit 154.60 before I could bring it under control (Lots of fun).

I do monitor Memory , CPU , DISK , CACHE and Network activity, but could not say that at ?X? value for a particular area I have a problem.

I tend to log in as a normal user and am aware of the ?Feel? of my machines.

Paula

If you can spell SysAdmin then you is one - anon

Hi
IMHO

It is very difficult to say what a threshold value should be as it varies for each server and its environment.

Each sysadmin gets a "Feel" for their machines.

It is normal to say that above 4.00 on top and the server is loaded - earlier this year I had to run a server that was averaging 12.00 - yes it was slower but we were still able to do business, it did runaway once and hit 154.60 before I could bring it under control (Lots of fun).

I do monitor Memory , CPU , DISK , CACHE and Network activity, but could not say that at ?X? value for a particular area I have a problem.

I tend to log in as a normal user and am aware of the ?Feel? of my machines.

Paula

If you can spell SysAdmin then you is one - anon

OOOPS - SORREEEEEE

The Sumbit seems to have problems - Or is it me and Monday morning?????

If you can spell SysAdmin then you is one - anon

Hi,

This is a large question : first, you could have a look at the help of SAM for kernel parameters.
You have to check first what kind of software is running on your N-Class, and the role of it.
Here are some "classical" kernel parameters that should be changed according to the use of your server :

maxuprc = Maximum Processes per user (>= 512 for example)
nproc = max number of processes on your server (> 3000 could be enough for example)
ninode = number of open inode (if processes use lots of files)
nfile = max number of open files (if large databases, you should increase this parameter for example)
maxusers = Maximum users connected to the server,
maxswapchunk = max number of chunks for swapping (in large RAM system)
dbc_min_pct = min number of buffer pages (Memory) for 1Gb RAM, 5% is enough
dbc_max_pct = max number of buffer pages
for 1 Gb RAM, 10% is enough
...

There are a lot...
The fact to increase thoses parameters just increase the size of the kernel in the memory, there is no risk to increase them...
So, to avoid limits, it is recommended to increase those parameters.

Hope it helps...

PJA

PJA