| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Beta Testing
Forum: Beta Test Support Forum
Thread: Server aborting CMD2 betas? [Resolved - working as expected] |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 9
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I recieved 7 CMD2 betas but 4 have since been Aborted by no action of my own. In my messages and in stdoutdae.txt it only says reporting completed tasks at the same time I see they were returned when I look at my Results stats. I think I see what happened now ... there were three initial replications and two returned and were validated before mine started which works from a testing standpoint (why crunch more than necessary) I didn't know this is how WCG is processing WUs. Can anyone confirm? Is this behavior limited to Beta in general, this run, or is it a beta of a general behavior WCG is investigating? I have seen this on other projects and I will say the learning curve for people to accept this behavior (they think it started crunching before it was aborted) appeared steep, the forums can get a bit tense and WCG is a much larger project so I would imagine the gnashing of teeth would be greater (at least in sheer volume) if this is the case.
----------------------------------------CMD2_ 0001-2R15_ B.clustersOccur-2YUX_ A.clustersOccur_ 6_ 0-- CMD2_ 0001-2R15_ B.clustersOccur-2YUX_ A.clustersOccur_ 1_ 1-- CMD2_ 0001-2R15_ B.clustersOccur-2YUW_ A.clustersOccur_ 4_ 1-- CMD2_ 0001-2R15_ B.clustersOccur-2YUW_ A.clustersOccur_ 13_ 0-- [Edit 1 times, last edit by Former Member at Apr 29, 2009 9:28:51 PM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
This is a function employed now for some time that allows to push a batch through quicker when need be yet have little waste, and no, a job started is not touched unless it's known to crash anyhow.
----------------------------------------Edit: Emphasis added
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Apr 29, 2009 9:20:13 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
This is a function employed now for some time ... and I thought I was pretty good at keeping up with things .. thanks for confirming.  [Edit 1 times, last edit by Former Member at Apr 29, 2009 9:28:35 PM] |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
The idea is to more clearly identify server side aborted results... maybe their status will be marked in future by something like "canceled by project/server". There's a few more Status identifier additions/changes in the pipeline such as Inconclusive maybe being called Pending Revalidation the former someone really getting a gnashfit over.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Renaming Inconclusive would hopefully result in the elimination of thousands of fora postings
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
maybe ... "canceled redundant result"
maybe ... "aborted redundant result" Maybe trying to combne staus and reason in just the status may start down a path ... status "aborted", the why "redundant" and eventually the who "client", "server", "project", "me" and this breaks down pretty quickly ... please summarize War and Peace in 16 charcaters or less  If not in the status itself perhaps a message could be added to the log with clarifications added for "who", "what", "why". Are there any thoughts on using this for production projects and not just beta? |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
It will be done so it adds value and removes the most obvious questions. The Pending validation is 2 liner. Inconclusive is 12 long, so 24 characters is allot to work with.
----------------------------------------I think you misunderstood. The feature is in place for all projects. If e.g. a No Reply copy is send out, the routine starts to monitors both the NR copy and the extra In Progress copy. Whichever comes back first, a subsequent signal is set ready for the client that has not returned the copy to cancel, if not yet started. Seen quite a few already with short deadline that subsequently got removed again because the NR copy came in sooner. The feature relies of course on the client talking to the servers before it starts on the extra copy. I know no statistics, but I think quite a few NR and extras are saved from crunching this way, without need. edit: some grammar
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Apr 30, 2009 10:50:30 AM] |
||
|
|
Flavio Bessa
Advanced Cruncher Brazil Joined: Aug 3, 2007 Post Count: 83 Status: Offline Project Badges: 
|
What about the credits? If someone starts working in a WU, then it gets aborted by server or something similar, at least it would be fair to grand partial credit to whoever got their WU aborted...
|
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Refer to my earlier comments. Jobs that have started, and do not have a known problem are left running, though redundant. There is no way to know if just started or near finishing, so they get the normal validation / credit treatment when completed.
----------------------------------------
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
|