| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Retired Forums
Forum: The New Members Forum [Read Only]
Thread: Need to disable device |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi, I have a device running with an agent folding proteins every day. But I need to have this client disconnected. How to i prevent the work from being lost? Can I somewhere tell the client to finish that task without receiving another one?
Does soemone know?? Thanks. Reiner |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The server will send out the Work Unit again eventually if you just uninstall. Here is something that I have never tried.
 Bring up the Agent interface, then bring up Task Manager. Use Task Manager to kill WCGrid_Rosetta. This should cause the Agent to connect to the server to report the error, then start to download a new Work Unit. Kill ud.exe before the transfer completes and is acknowledged. Then uninstall. This ought to leave the server knowing that your device is not running anything.Yes, I think this is a kludge too.  But I cannot think of anything better. Comments, anyone?Lawrence |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The server will send out the Work Unit again eventually if you just uninstall. Here is something that I have never tried. Bring up the Agent interface, then bring up Task Manager. Use Task Manager to kill WCGrid_Rosetta. This should cause the Agent to connect to the server to report the error, then start to download a new Work Unit. Kill ud.exe before the transfer completes and is acknowledged. Then uninstall. This ought to leave the server knowing that your device is not running anything.Yes, I think this is a kludge too. But I cannot think of anything better. Comments, anyone?Lawrence Hi Lawrence You can right click on the WCG Agent Icon in the bottom right hand side of the task bar and Click on Exit without losing the progress of the current work unit Then just click on Start/Programs/World CommunityGrid/ World Community Grid Agent and restart it when you are ready and it will progress from where it left off Regards [Edit 1 times, last edit by Former Member at Feb 8, 2005 10:36:51 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Graham, I think that you are right. I think I misunderstood what he wanted to do.
Lawrence |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
You can right click on the WCG Agent Icon in the bottom right hand side of the task bar and Click on Exit without losing the progress of the current work unit Then just click on Start/Programs/World CommunityGrid/ World Community Grid Agent and restart it when you are ready and it will progress from where it left off Graham, Are you sure that this is exactly true ? Rosetta uses checkpoints at various points to write the current progress to disk. If you are in a big crunch and a checkpoint has not been written, then you will lose the crunching time between checkpoints will you not ? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Yes, a shutdown, whether using 'Exit' or a Log Off, does lose the computation since the last check point. But checkpoints are usually only a few minutes apart.
|
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
But checkpoints are usually only a few minutes apart. Whilst I agree that generally this is the case, here is an entry from my UDmon log: 09/02/2005 15:30:58: #1: Save (+ 0:01:31) which basically says that it was over an hour and a half between saves. That would be a lot of crunching time to lose. I would recommend that people goes to the application screen (the i logo) of the agent and only terminate after it has gone to a full percent (ie 10%, 11% etc) as this is when a checkpoint happens (at least for me). |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Stuart,
As far as I know, the checkpoints are written when the tenths of a percent change. That is why they jump several tenths at a time, instead of smoothly increasing. The program calculates how much progress has been made when it writes the checkpoint. This comes from a post by Rick Alther back in early December 2004, I think. Lawrence |
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
But checkpoints are usually only a few minutes apart. Whilst I agree that generally this is the case, here is an entry from my UDmon log: 09/02/2005 15:30:58: #1: Save (+ 0:01:31) which basically says that it was over an hour and a half between saves. That would be a lot of crunching time to lose. I would recommend that people goes to the application screen (the i logo) of the agent and only terminate after it has gone to a full percent (ie 10%, 11% etc) as this is when a checkpoint happens (at least for me). But I need to have this client disconnected. Hi Stuart If he absolutely has to shut it off because of an overiding problem, is it not better to lose 1.5 hours than kill it and lose the lot I believe RKnudsen priorities would overide WCG Regards [Edit 3 times, last edit by Former Member at Feb 9, 2005 7:58:57 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Stuart, As far as I know, the checkpoints are written when the tenths of a percent change. That is why they jump several tenths at a time, instead of smoothly increasing. The program calculates how much progress has been made when it writes the checkpoint. This comes from a post by Rick Alther back in early December 2004, I think. Lawrence Lawrence, That is indeed what Rick said. If you're looking to shut down the Agent without losing work, go to the (I) screen (with the protein graphic on it) and wait until it bumps up to the next tenth of a percent. That change indicates a checkpoint. The time it takes to get to the next checkpoint will depend (like all things Grid) on how fast your computer is, what else you're running, etc... Nels |
||
|
|
|