perfSONAR User Q&A and Other Discussion

[Jason Zurawski <>]

Hi Jim;

I am going to CC the Performance Node User list on this response as well.


> I've got a nice dell 1850 that I'd like to use to try out the toolkit. 
> It has a single 70 gig
> drive.  I see the recommendations, but I don't see minimums.  I tried to 
> see if I might
> find helpful prompts in the install procedures.  There is nothing to 
> tell me how many
> partitions I need and what their minimum sizes need to be, other than 
> the example
> that it might complain about 8 GB.


There are only two soft requirements for the pSPT:

 - 512M RAM
 - 10G of storage

These are soft because if they are not met the toolkit will still 
function, but warnings will appear because these are too low to be 
useful for long term measurement activities.

Only a single ext3 partition (10G or better) is necessary.  It is 
possible to use set up more partitions, but the common use case is to 
devote the entire disk for the pSPT.


> Is this hopeless?
>
> What I'd really like to see is an install script that told me that I 
> have an XX size
> disk drive and therefore here's what we recommend.  OK (Y/N)?
> This will wipe your disk drive, OK (Y/N)?


The current menus are designed to:

 - List the available partitions on a disk and their sizes.  It is up 
to the user to try to make a good choice in this case.
 - If the selected partition is not ext3, it will need to be formated.
 - If the selected partition is ext3, an option is given to format or 
use the disk with existing contents preserved.

I will make a note in our tracker that the menus are not completely 
clear on the options and also edit 
http://code.google.com/p/perfsonar-ps/wiki/pSPerformanceToolkit31#Storage 
to explain what is happening in a more verbose way.


> I did go into the custom formatting screen and that seems to be just a 
> disaster.
> On my KB, the arrow keys seem erratic.  The mouse doesn't work, and the
> toolkit provides no instructions in any case.


Editing the partitions is done though nparted.  I am not sure if we are 
running gpm (I would doubt it) so the mouse not working should be 
expected.  We will try to reproduce the other behavior you are seeing, 
but since this is coming from an external application there may not be 
much we can do.

Thanks;

-jason