perfsonar development work

[Fausto Vetter <>]

Hi,

Few comments.. :-)

Jochen Reinwand wrote:
> - Hades is divided into a central server and the measurement boxes. Hades MA 
> is only running on the central server and the central server has a complete 
> overview of the configured and running measurement. Features missing (like 
> time synchronisation accuracy) are planed to be added to the measurement 
> information so that the central server knows "everything".
>   
This clock information is already predicted in the hardware layer.. it 
should be the same for all services... I think it is a requirement for 
measurements at a whole.
> - Status information from the measurement boxes that is still important, is 
> nearly completely covered by the "Hardware Layer monitoring".
>   
Good. So we will monitor the same for all servers.
> - For Hades and BWCTL some special things from the "Hardware Layer" should be 
> monitored, though. For reliable measurements both need a good time 
> synchronisation and a tuned network interface. I'm not sure how to insert 
> these requirements into the layer schema, because I see them as normally 
> related to the Hardware Layer, but specific for the corresponding Service 
> Layer. Also it is important that neither the needed NTP synchronisation 
> accuracy nor the tuning of the network interface are the same for BWCTL and 
> Hades.
>   
Put this service related, because this is a resource being used by the 
service. If you think general, you see it is not true for all services 
across.

By the way, how to monitor hades central box? no specific metric was 
defined on the wiki.... :-)

Fausto