Re: oM 2.6 signals and multiple front nodes.

Hi.

I think you are missing my point. I am well aware that a job can be 
started from any node, thats why I would like to use LVS. It would seem 
to the users (which are not very tech savvy) that they are logging into 
the "cluster" and would solve the problems i discussed (at least 
partially). The problem is that I can't (easily) monitor activity on the 
cluster.

Openmosixprocs almost solves the problem, but not quite. It allows me to 
see processes started on other nodes IF they are running on the same 
node as openmosixprocs. What I want is to see the users processes, no 
matter which node they logged into. Maybe I have missed some feature of 
openmosixview that lets me do that.

It is also necessary for the users to be able to log in at different 
points in time and from different locations and still be able to see 
their processes, which isn't possible today unless they log in on the 
same node all the time.

Regards, Pontus.

Bruce Knox wrote:

>Pontus, take a look at openMosix 2.4.26-1 with openMosixview+ssh and CHeckPOinting for linuX, CHPOX.
>openMosixview+ssh will give you the ability to see and manage the processes on all the openMosix nodes.  Both can be found on the Community page http://openmosix.sourceforge.net/community.html
>
>As Evan pointed out, openMosix will allow any node to start a process.  Even though many openMosix users think of their cluster as having a head node, openMosix really does not have a head node.
>
>CHPOX is extremely useful.  Most users would think of it for protecting your cluster by minimizing losses due interruptions such as crashes, power failures, etc.  
>
>But with a little effort, I think you will find that you can Checkpoint a process and actually move the Checkpointed process to a different node (or even another similar cluster).  
>
>Bruce
>
>  
>
>>>>Pontus Pihlgren <pontus.pihlgren-dMXaqZTro8s4M/[email protected]> 12/02/05 04:42AM >>>
>>>>        
>>>>
>Hi.
>.
>I had two questions about openMosix 2.6, the first was about signals. It 
>seems signals are a bit busted in oM 2.4, but a quick search in the 
>mailing list archives reveals that (according to Moshe) signal are 
>handled in a new way in 2.6. Would this new implementation enable us to 
>SIGSTOP lower priority jobs(on remote nodes) in favour of those with 
>higher priority and later SIGCONT them?
>
>My second question, which I couldn't find any information about, is that 
>of multiple front nodes. I would very much like to have multiple front 
>nodes for our cluster. As it is now, we are very dependent on the front 
>node. If it crashes, everything crashes. It is also very hard to do 
>maintenance and updates to the front node, since it is active almost 
>24/7. Something like Linux Virtual Server (LVS) would solve that, but 
>makes it very hard to get an overview of the status of the cluster since 
>processes started on one node will not be seen in the process lists of 
>the others.
>The real question then, will 2.6 enable one node to see processes 
>started on another node?
>Of course not all processes would be of interrest (such as deamons and 
>the like), but at least the processes that are not locked? or the 
>processes from userspace? One could think of a range of filters for 
>which processes to show, but that might be left to userland tools to handle.
>
>If it would be possible, it would be a big reason for us to switch to 
>2.6 once the userland tools are stable. For now, 2.4 is working almost 
>flawlessly. Thank you all.
>
>Regards,
>Pontus.
>
>
>-------------------------------------------------------
>This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
>for problems?  Stop!  Download the new AJAX search engine that makes
>searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
>http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click 
>_______________________________________________
>openMosix-general mailing list
>openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org 
>https://lists.sourceforge.net/lists/listinfo/openmosix-general
>
>
>
>-------------------------------------------------------
>This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
>for problems?  Stop!  Download the new AJAX search engine that makes
>searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
>http://ads.osdn.com/?ad_idv37&alloc_id865&op=click
>_______________________________________________
>openMosix-general mailing list
>openMosix-general-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f@public.gmane.org
>https://lists.sourceforge.net/lists/listinfo/openmosix-general
>  
>



-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click