-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Implementing my patches would lower the load some ( only sends data to the
master that is going to the alarm backend ).

This lightens the load some, but on our enviroment we are having load problems
because of all the http connections being made from the satellites to the
master ( load average higher than 30, peak at 600 :( )

We are currently investisgating the posibility to not send any data to the
master at all, but let the master ask for everything on demand ( will most
probably lighten the load considarbly ).

Our temporary fix for the load problem is to serialize all the data that is
supposed to be sent to the master, and put into a table ( one row for each 
call )

And then have a script that is reading this table every 10th seconds and send
all the data in one batch, this has lowered the load on the master to
acceptable levels, but have risen the load on the satellites from 2 to 30
( polling 1800 interfaces at the moment )

If anyone is interested in our temporary path for the backend ( requires the
earlier patches to implement a framework for calling a backend ), i can post
it to the list.

Cheers
Anders

On Sunday 06 June 2004 02.06, Javier Szyszlican wrote:
> Hi,
>
> Satellites are what you think, but they don't really lower the load,
> because the Master still needs to do the 'backend' part of the polling,
> only the pollers are distributed.
>
> All data is stored in the master. So, it still have around the same load.
>
> This is useful in situations where the master can't access the remote
> network directly, or when you want 2 pollers polling the same host in
> case some of them die.
>
> I expect that in the future the satellite system  will case lighter loads.
>
> Another option is implementing some patches sent to me by Anders
> Karlsson that make the RRD Storage and Graphing distributed too, so the
> master will ask the childs (slaves) for graphs, but the polling is local
> to the slave.
>
> Javier
>
> Tjerk Nan wrote:
> > Hello,
> >
> > As i'm getting familiar with JFFNMS i discovered a potentialy
> > interesting feature: satelites. Although I can't find any information
> > about how I should use it, I guess you can use it to distribute the load
> > JFFNMS is generating (which can be high!) on more machines. Can somebody
> > tell me more about the principle behind satelites?
> >
> > With regards,
> >
> > Tjerk Nan
> >
> >
> >
> > -------------------------------------------------------
> > This SF.Net email is sponsored by the new InstallShield X.
> >  From Windows to Linux, servers to mobile, InstallShield X is the one
> > installation-authoring solution that does it all. Learn more and
> > evaluate today! http://www.installshield.com/Dev2Dev/0504
> > _______________________________________________
> > jffnms-users mailing list
> > [EMAIL PROTECTED]
> > https://lists.sourceforge.net/lists/listinfo/jffnms-users

- -- 
SongNetworks
[EMAIL PROTECTED]
CDTT (Certified Duct Tape Technician)

Programming today is a race between software engineers striving to build 
bigger and better idiot-proof programs, and the Universe trying to produce 
bigger and better idiots. So far, the Universe is winning. - Rich Cook
 

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (FreeBSD)

iD8DBQFAxXGDi30A6pYsr0wRAoAgAKC0tYda3Lq+Lt+IqrR1qpUn5YH17wCfXgfR
FfbgyZmLp1rs2X9vsQDayHM=
=4iY8
-----END PGP SIGNATURE-----


-------------------------------------------------------
This SF.Net email is sponsored by: GNOME Foundation
Hackers Unite!  GUADEC: The world's #1 Open Source Desktop Event.
GNOME Users and Developers European Conference, 28-30th June in Norway
http://2004/guadec.org
_______________________________________________
jffnms-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/jffnms-users

Reply via email to