ORBIT-USER: Question about Virtual Mobility Tutorial
Arinto Murdopo
arinto at gmail.com
Wed Feb 21 08:49:40 EST 2007
Hi Ivan,
Does it mean that nodes are more reliable after maintenance period? I found
some weird things, the node is displayed as active and on in the ORBIT
webpage, however, I cannot transfer file to the node, it said no connection
to the nodes although I have switched on the specified node.
Thank you,
Arinto :D
On 2/21/07, Ivan Seskar <Seskar at winlab.rutgers.edu> wrote:
>
> Hi Arinto,
>
> What you observed is a power supply problem in action. There is not much
> you can do about it so just please ignore it; we will try to fix some of
> them during the maintenance slot.
>
> Ivan.
>
> -----Original Message-----
> From: owner-orbit-user at winlab.rutgers.edu [mailto:
> owner-orbit-user at winlab.rutgers.edu] On Behalf Of Arinto Murdopo
> Sent: Wednesday, February 21, 2007 4:46 AM
> To: orbit-user at winlab.rutgers.edu
> Subject: Re: ORBIT-USER: Question about Virtual Mobility Tutorial
>
> Dear all,
>
> I have tried the allOffSoft command to switch off all nodes inside the
> main grid. However, when I view the status via
> http://www.orbit-lab.org/wiki/Status/Grid , some nodes still in the on
> state. I tried to repeat the command for 3 to 4 times every 2 to 3 minutes,
> but the result is still the same T_T.. Same nodes still in the on state.
>
> What are the possible causes for this problem? I remembered few days ago I
> was able to switch off almost all nodes (only left nodes that are not
> available) using this offSoftCommand. Could it be someone still log in to
> the nodes?
>
> Thank you
>
> arinto
>
>
> On 2/21/07, Arinto Murdopo <arinto at gmail.com> wrote:
>
> Okie thank you.. yup, like what you stated above, sometimes there
> are few nodes that still in on state although I have issued allOffSoft.
>
> Thank you :D
>
> arinto
>
>
>
> On 2/21/07, Sanjit Krishnan Kaul <sanjitkaul at gmail.com> wrote:
>
> What I meant was synchronizing the installation of click
> scripts across nodes that are part of your experiment. So it could also mean
> synchronization between the sender and the receiver but only as far as
> installation of click scripts is concerned.
>
>
>
> You could try integrating shell scripts that load the
> click scripts or you can have your own mechanism.
>
>
>
> allOffSoft should take 1 - 2 mins max. Fire the command
> again in case you see some nodes that have not switched off.
>
>
>
>
> ________________________________
>
>
> From: owner- orbit-user at winlab.rutgers.edu <mailto:
> orbit-user at winlab.rutgers.edu> [mailto:owner-
> orbit-user at winlab.rutgers.edu <mailto:orbit-user at winlab.rutgers.edu> ] On
> Behalf Of Arinto Murdopo
> Sent: Tuesday, February 20, 2007 6:23 AM
> To: orbit-user at winlab.rutgers.edu
> Subject: Re: ORBIT-USER: Question about Virtual Mobility
> Tutorial
>
>
>
> Okay, thank you.
>
> By the way, what do you mean by synchronization? Does it
> mean synchronization between sender and receiver?
>
> I read from one of ORBIT publication that nodehandler
> application is used for synchronization. In this case, I think I need to use
> integrate the shell scripting codes (something like loading file, set the
> file status etc) into ruby script. Is that correct? Please correct me if I'm
> wrong.
>
> By the way, one more question, how long does it take to
> complete the process of switching off all nodes in the main grid using
> allOffSoft in CMC command?
> Thank you
>
> arinto
>
> On 2/20/07, Sanjit Krishnan Kaul <sanjitkaul at gmail.com>
> wrote:
>
> Your understanding is mostly correct. A specific node will
> start switching as soon as the node's click script is installed on it (using
> the command click-install). By "switching" I mean processing packets as
> defined by the node's click script.
>
>
>
> Nodes don't wait for other nodes to install their click
> scripts. Incase you want them to start switching at the same time you have
> to ensure synchronization between the nodes.
>
>
>
> Hope that helps!
>
>
>
> PS: The command in the example says click-install
> mob*.click. The "*" means the click script downloaded to the node where you
> are running the click-install command and not all scripts. Also, in general
> you shouldn't install more than a click script on a node. If you do the last
> installed script will determine the switching rules the node follows and the
> rest will be ignored.
>
>
>
> Sanjit
>
>
> ________________________________
>
>
> From: owner- orbit-user at winlab.rutgers.edu <mailto:
> orbit-user at winlab.rutgers.edu> [mailto:owner-
> orbit-user at winlab.rutgers.edu <mailto:orbit-user at winlab.rutgers.edu> ] On
> Behalf Of Arinto Murdopo
> Sent: Tuesday, February 20, 2007 3:11 AM
> To: orbit-user at winlab.rutgers.edu
> Subject: ORBIT-USER: Question about Virtual Mobility
> Tutorial
>
>
>
> Dear all,
>
> I have question regarding the Virtual Mobility Tutorial
> using spatial switching (
> http://www.orbit-lab.org/wiki/HowTo/virtualMobilitySS). Last statement of
> the last section as shown below:
>
>
> Running an experiment ¶ <
> http://www.orbit-lab.org/wiki/HowTo/virtualMobilitySS#Runninganexperiment>
>
>
> * Image all grid nodes (VM nodes and the nodes
> mapped to mobile nodes) with the image click-2.4.26-20051204.ndz.
> * Upload the click files onto the grid nodes they
> correspond. The file loadFiles.sh contains sample commands for the example
> experiment.
> * For each node install the click script. On nodes
> specified in the VM<no>.txt(s) it is important that the ath0 interface is
> up.
>
> $ ssh root at nodex-y
>
>
>
>
> $ click-install mob*.click
>
> * The installation of the click script marks the
> beginning of experiment on each node.
>
> What does the last statement mean? Does it mean that the
> node start switching when we finish the installation all the click
> configuration files using click-install?
> When is the time actually the specific nodes start
> switching?
>
> Thank you
>
> Arinto
>
>
>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://orbit-lab.org/pipermail/orbit-user/attachments/20070221/4c4b3097/attachment.htm
More information about the orbit-user
mailing list