| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: Suggestions / Feedback
Thread: Restartibilty |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 8
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hi,
I am using UD (for windows XP) and find very awkward regarding following issues: 1. You cannot start it explicitly I mean, if my system goes is power efficient mode then UD goes into sleep mode and i cannot find way to activate the same. Other proceses in my system work perfectly well, in power efficient mode. 2. Poor restartibility Many times, it showed a task completed 80% and sometimes 90% too. and if UD crashed, it will start from scrath again. THIS IS A BUG!! can someone fix this? I mean how can you afford to loose 80% of data on an application crash , that took 5 hrs to compute. i think it needs immediate addressing. (By the way, crash was due to some other program running on system and not of UD) |
|||
|
||||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
On 1. UD Agent and BOINC are trained to not process unless the computer is not in energy saving mode / running on battery / running on UPS backup.
----------------------------------------On 2. Strange as no- or few segmented FAAHs (Fighting Aids At Home guess?), have been in circulation for a long time. Can only think of true corruption of the intermediate result file at which time the Work Unit is restarted from scratch....any system crash can cause such an event. Questions: Is it FAAH or Help Defeat Cancer U R talking about.... u did not post in the relevant project thread. ciao
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 3 times, last edit by Sekerob at Oct 5, 2006 5:48:47 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
1. My system is "on" for some overnight builds and stuffs .. and generally goes into power efficient mode (which is not equivalent to standby ..!). As such, one should be allowed to run UD, explicitly.
I dont know how much reasonable it sounds to you :)) 2. if there was no checkpoint done at 80%, then its worth considering. loosing 5% of data makes sense in case of crash, but loosing full 80% .. nahhhh .. !! anyway, i will ping again if it happens again. is there any log file generated? that i can analyse or sent across. that reminds me of one more feature that can be put into the application, i.e. report any bug directly from the application. thanks for the reply! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
oops i forgot to mention ..
thread is for FAAH !! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
UD should not stop if your system goes into a low-power mode. However, there are many different power saving modes, and it is impossible to predict UD's behaviour under all conditions. You will have to tell us more if you need further advice on this.
Work units have different numbers of checkpoints. Ideally, they checkpoint very frequently. There are a rare few that don't. There is no log data, UD is proprietary and works as a black box on the client. If you want more control, switch to BOINC. It's much better. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Ritesh
----------------------------------------I'm fairly new to WCG but have a little experience of Distributed Computing. In regard to the data loss in case of a crash situation, I have seen many posts on many forums about this type of thing and do not know of any projects that have been able to overcome this problem. Folding@Home has the same problem and they write checkpoint files at as little as Three minute intervals. However if the client is stopped incorrectly i.e. a crash all of the work(up to 2 weeks or more worth) will be unrecoverable. I don't know why this happens but I do know it has been worked on by many excellent programmers and technicians without ever being solved. I think we just have to live with it in this type of project. Cheers. ozylynx.  [Edit 1 times, last edit by Former Member at Oct 5, 2006 7:41:21 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
ozylynx, I believe you are correct. BOINC can recover from the last checkpoint in the case of a crash, UD cannot.
Whether this is because UD can't, or whether it just plays safe, I do not know. Anyway, a good reason for shutting down properly. If an application crashes, then kill that application - if the operating system crashes, then reinstall the operating system! And if you're using Windows 98, stop. ;-) |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As long as the check point file is not corrupted by a crash, either BOINC or UD will start at the last check point. If the check point file is open and being written to when the crash occurs, then the work has to start at the beginning. Different work units write check points at different times. A very few FAAH work units never check point. It all depends on the path taken through the program.
Lawrence |
||
|
|
|