Hi Andy,
If we need something on the client side, I agree completely with you.
I'd like to see if client side monitoring is needed or if the server
side accomplishes everything that is really needed. Doing it on the
server side has a number of advantages, I think. Sounds like we can work
through this in the xrootd meeting on 18 Oct.
Pete
On Thu, Oct 06, 2005 at 11:52:16AM -0700, Andy Hanushevsky wrote:
> Hi All,
>
> I think the best (though most labor intensive approach) is to define a
> plug-in architecture that would accomodate any kind of monitoring. Then
> it's just a matter of a config file or even, perhaps, command line option.
>
> Andy
>
> ----- Original Message -----
> From: "Derek Feichtinger" <[log in to unmask]>
> To: "Peter Elmer" <[log in to unmask]>
> Cc: "Fabrizio Furano" <[log in to unmask]>; "xrootd-list"
> <[log in to unmask]>; "Catalin Cirstoiu" <[log in to unmask]>
> Sent: Thursday, October 06, 2005 5:38 AM
> Subject: Re: xrdcp with MonALISA monitoring
>
>
> >Hi, Fabrizio and Peter
> >
> >You are right. Let's do it like this: We leave this completely out for
> >now.
> >I will generate a patch to xrdcp which will be used in the Alice builds.
> >
> >After talking to Catalin I can see now that the code is not in a state
> >that it
> >should be added to the common CVS anyhow (I had suspected a bit, an that's
> >why I did not ask for inclusion of the #ifdefs in the present xrdcp).
> >Catalin
> >will get in contact with you again whether it is possible to get all the
> >information he collects via the client extension also via the server based
> >monitoring (added Catalin in CC).
> >If it turns out that there is interest in this functionality, we can still
> >make it a configure option.
> >
> >Cheers,
> >Derek
> >
> >
> >On Thursday 06 October 2005 13.45, Peter Elmer wrote:
> >> Hi Fabrizio and Derek,
> >>
> >> I'm also not fond of the idea of a separate executable as in any case
> >>then we are at the mercy of people using the correct one from any given
> >>installation of xrootd. It would be somewhat cleaner if people always
> >>just used 'xrd', whereever they are, and _if_ the command was linked
> >>with MonaLisa monitoring support it uses it. I guess there is then the
> >>question of how it knows whether and where to send the information (if
> >>the
> >>support is linked). How would that have worked with the 'xrdmoncp'
> >>command?
> >>
> >> BTW, I'm still not sure what the _client_ side support for MonALISA
> >>gives us that that the server side support would not (e.g. by reusing
> >>the existing monitoring and simply allowing it to be redirected to
> >>MonaLISA). Could you spell this out for me, Derek?
> >>
> >> Pete
> >>
> >>On Thu, Oct 06, 2005 at 01:35:38PM +0200, Fabrizio Furano wrote:
> >>> Hi Derek,
> >>>
> >>> generally I don't like "special" executable targets. However, I have
> >>> no
> >>> objections if the idea is to move to a unique way when those libraries
> >>> are available.
> >>>
> >>> When I am out from the boring stuff I am on (very soon I hope) I will
> >>> start with the developing of some new features. Obviously I'll not
> >>> replicate modifications onto different code branches...
> >>>
> >>> Fabrizio
> >>>
> >>> On Wed, 2005-10-05 at 16:29 +0200, Derek Feichtinger wrote:
> >>> > Ciao, Fabrizio
> >>> >
> >>> > We are using here a xrdcp with some added code for MonALISA
> >>> > monitoring
> >>> > of the clients. It's just the normal client with quite some stuff in
> >>> > between #ifdefs and it links against a MonALISA client library
> >>> > (libapmoncpp). The library is unluckily snot yet available on the
> >>> > main
> >>> > site (no official download page for that version yet. Catalin said he
> >>> > will include all the necessary changes in the official MonALISA
> >>> > soon).
> >>> >
> >>> > For these reasons I would like to include it for now as a separate
> >>> > second client called xrdmoncp in the XrdClients directory, that will
> >>> > only be built when autotools detect the presence of the libapmoncpp
> >>> > library. This way it's easy for users to detect whether a copy
> >>> > command
> >>> > with MonALISA monitoring is available.
> >>> >
> >>> > Is this ok for you?
> >>> >
> >>> > Any others interested in this monitoring possibility?
> >>> >
> >>> > Cheers,
> >>> > Derek
> >>
> >>-------------------------------------------------------------------------
> >>Peter Elmer E-mail: [log in to unmask] Phone: +41 (22) 767-4644
> >>Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
> >>-------------------------------------------------------------------------
> >
> >--
> >Dr. Derek Feichtinger Tel: +41 22 767 10 07
> >ARDA project email: [log in to unmask]
> >CERN
> >http://people.web.psi.ch/feichtinger
> >CH-1211 Genève 23
> >
> >
> >
-------------------------------------------------------------------------
Peter Elmer E-mail: [log in to unmask] Phone: +41 (22) 767-4644
Address: CERN Division PPE, Bat. 32 2C-14, CH-1211 Geneva 23, Switzerland
-------------------------------------------------------------------------
