| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Trusted Computer? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 32
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Knock yourself out: http://www.trustmanagement.org/2010/06/trusted-computers-are-reliable-but.html
--//-- |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sek, how could you have missed my "... in the context of WCG's server's assessment of client machines connecting to it"?
 ; |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges: 
|
Is "trusted", and "reliable" as used in the context of WCG's server's assessment of client machines connecting to it -- mean the same thing? I recall there is a difference. ; No they're different, since "trusted" goes on unreplicated work in cases project uses "adaptive replication", while "reliable" goes on fast turnaround-time then a wu is re-issued for any reason. The limits for how many Validated in a row can also be different between "trusted" and "reliable", but it's also possible WCG uses the same limits. So, if assumes it's the same limit, it's like this: A computer is "Trusted" if: 1: The per-application, per-plan-class consecutive_valid >= 5. A computer is "reliable" if: 2: It's "Trusted" for the per-application, per-plan-class. 3: The computers total daily quota is 120 per core. 4: The average turnaround-time < 2 days. In case it's a non-intel Mac, the limit is < 3.6 days.  "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Ingleside.
Reference: Ingleside [Apr 27, 2012 1:15:11 PM] post Excellent response!  I thought there was a need to first clarify the difference between "trusted" and "reliable" for, as I was reading the earlier posts, I'm not sure if folks are clear about the said difference. I'm not sure myself, and most everyone mentioned "reliable" whereas the OP title is about (questions regarding) a "Trusted" computer. Anyways, your response certainly made the picture clearer. I'd like to see the Techs chime in on this one so that everyone is on the same page about the numbers for the "reliability" or "trusted" computers. ; |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Go back to first reply and read again, reproduced here for your convenience. What's not clear about "reliable" as explained by knreed and split to 3 points in my expansion or the qualification given of Trusted (only 5 "consecutive" valid returns for a science app) by Ingleside?
----------------------------------------While we are discussing, for a host to be reliable, this again is now done at the app_version level. To be reliable a host must have 5 consecutive valid results, be allowed the full daily quota of results for the app_version (120), and have a average turnaround for the app_version less than 2 days. A. Reliability is no longer based on a device basis, but on a device/science app combo. B. The device must have met 3 criteria for a specific science. 1. Done 5 Valid results consecutively (not returned, but validated). 2. The maximum quota must be applying (120 per core per day). Errors reduce that quota immediately to a lower number. 3. Your results must get reported (the ready to report bit), on average within 48 hours. --//-- P.S. Being reliable does not mean you get them. Depends on how many reliable devices hav been rated for each of the individual sciences, and how many errors need to be fixed... if few, you wont see them either or rarely. ATM I'm getting them frequently on octo and quad for CFSW, but at 60-70 per day without error, that means loads of validation and re-confirmation the hosts are fit to be in the ''second opinion'' league. Whilst, long moons ago knreed was suggesting a congratulations message to reliable hosts (we did not discuss a retraction message ;P)... who knows one day, one spare hour and we'll get it... it's one of those ''nice to have, but not essential''. [Edit 1 times, last edit by Former Member at Apr 27, 2012 5:13:01 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Go back to first reply and read again... Ok, done. I re-read it. Result: not even one word there that spells t, r, u, s, and t -- or "trust". Also, please take note of the OP title: Trusted computer? and not Reliable computer? Number of words that spell "reliable" in your post: 5. So were you talking about trust? I don't think so. You were talking about reliability.And my original question, precisely with the motivation to clear the air about the possible confusion between "trusted" and "reliable" -- reproduced below for your convenience, was: Is "trusted", and "reliable" as used in the context of WCG's server's assessment of client machines connecting to it -- mean the same thing? I recall there is a difference. ; |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges: 
|
A computer is "Trusted" if: 1: The per-application, per-plan-class consecutive_valid >= 5. Thanks for the clear explanation. I should be trusted, and do get a few in each project that have short fuses while the majority have longer (10 day) time limits. So I think they are not yet consistent in implementing their plan, but hopefully it will be cleared up eventually. EDIT: Or maybe it is the per-plan-class restriction that has to be met frequently? Whatever they want. [Edit 1 times, last edit by Jim1348 at Apr 28, 2012 9:18:26 AM] |
||
|
|
Ingleside
Veteran Cruncher Norway Joined: Nov 19, 2005 Post Count: 974 Status: Offline Project Badges:
|
Thanks for the clear explanation. I should be trusted, and do get a few in each project that have short fuses while the majority have longer (10 day) time limits. So I think they are not yet consistent in implementing their plan, but hopefully it will be cleared up eventually. Actually, the majority of your work having long deadline is a good thing, since you'll only get a shorter deadline if someone else: 1: Didn't manage to download their task. 2: Crashed on their task. 3: Didn't return their task by the deadline. 4: Aborted their task. 5: Returned their task, but it failed validation. Meaning, the more short-deadline tasks you're getting, the higher error-rate other users are having. The higher the error-rate is, especially if it's due to #5, the more wasted resources there is. So, if you never get any short-deadline work, even your computer is "reliable" in all the projects, would actually be a good thing, since this would indicate very low error-rate. Making a very easy example, let's assume all work-distribution and so on is uniformly distributed. Let's only look on a single application/plan-class, and let's assume 25% of all computers is "reliable" for this. On a "reliable" computer, if 100% of downloaded work has a short deadline, this means at least 25% error-rate and upto 75% error-rate, something that would be really bad. On a "reliable" computer, if 25% of downloaded work has a short deadline, this would indicate 6.25% error-rate. On a "reliable" computer, if 10% of downloaded work has a short deadline, this would indicate 2.5% error-rate. On a "reliable" computer, if 1% of downloaded work has a short deadline, this would indicate 0.25% error-rate. Personally, I would much rather my computers is at 1% short-deadline work than on 10%. "I make so many mistakes. But then just think of all the mistakes I don't make, although I might." |
||
|
|
Jim1348
Veteran Cruncher USA Joined: Jul 13, 2009 Post Count: 1066 Status: Offline Project Badges:
|
So, if you never get any short-deadline work, even your computer is "reliable" in all the projects, would actually be a good thing, since this would indicate very low error-rate. Fine. I thought that being "trusted" meant that a work unit did not need to be verified by someone else. I guess that is a different thing. |
||
|
|
BudWheel7
Cruncher Joined: Mar 24, 2012 Post Count: 1 Status: Offline |
I am running 10 Desktops 24/7. Where does one get all info about how Boin operates. Not a techie, just plug, download and play. Do not know about pick of litter machines, etc.. Just know I average 50-60 PV's per PC after daily evening processsing. If there were less pending verifications wouldn't need as many machines. Who is good contact to answer questions ?
|
||
|
|
|