| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 26
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Progress! Now we can look through Viktors' list to try to identify the problem that would cause the crashes. I think that the easiest one to check is the Virtual Memory. mycrofth |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Wayndal --
Rick or Viktors may want to ring in on your second item, but my guess (and it is a guess because I don't have access to the actual data that they do) is that you may have had what I have seen refered to a "monster work unit" complete on that day prior to going into your blocked shutdown state. Around the 1st of the year, several of us on the team I am a member of (under my "real" member name) noticed that several of these monsters were running. Depending on the PC, they were taking 20 hours up to 2 1/2 to 3 days to complete. Since the points and runtime are associated with the date the unit completed, it is likely you had been crunching one of these and it completed sometime that day before you performed your shutdown. Regards, |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Mycrofth --
It would seem to me that a system which stuck while going through shutdown is an unpredictable environment and not one likely to be considered a supported environment by the project team. That's just my take on it. |
||
|
|
wayndal
Cruncher Joined: Nov 23, 2005 Post Count: 17 Status: Offline Project Badges:
|
Hmmm,... if anyone cares, I've got 1536 MB of virtual memory. Not sure how to check for leaks though.
- If I WAS running a "Monster", wouldn't my results returned be low and my points higher? Because all 3, results returned, runtime and points had a significant increase (about 2.5x "normal"). - I think though, that I'd have to agree with LC,... in such a "freak" arrangment as a jammed shutdown (although it isn't THAT messed up...) may not be worth too much research. On the other hand, there was that one day (see my last post) that may show some kind of light for it? And I sure don't mind. You may have to explain to me how to get some of the numbers and info that you may want, but I sure don't mind telling you guys whatever you want to know about it. Lastly, unless I hear otherwise, my PC will be in "normal mode" over the weekend. If you want me to change it, let me know ASAP. Thanx!! ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Wayndal --
- If I WAS running a "Monster", wouldn't my results returned be low and my points higher? Because all 3, results returned, runtime and points had a significant increase (about 2.5x "normal"). The reason I was guessing that you had the "monster" work unit was your earlier statement: On a date around ( I think ) Jan 1st, it looked as though the results returned spiked but so did my points and runtime. Does this mean that the results were completed, or that possibly the points were mis-calculated? Or some other reason? Not a big deal. Just the only thing that keeps your explination from completly answering everything. If one of these had finished earlier in the day before you started the shutdown that would not complete because of ICQ, the runtime and points for the "monster" would accrue to you that day. Say, for example, that work unit had taken 36 hours to complete and was worth perhaps 1200 points and that it completed at 03:30 GMT that day. Then, another work unit would have started and perhaps it got up to 40% completion when you entered shutdown at maybe 06:30 GMT. From what Viktors was seeing, I am assuming that all the work units you are seeing counted in the results from the shutdown period are crashing and are only being added to the results counter and are not adding to the run time or to the points and that they are crashing on an hourly basis. If the one that was at 40% crashed immediately and once an hour thereafter a new work unit was obtained that crashed, this would add 18 work units and no run time or points to the day's totals. Thus, for that day on that device, you would have seen: Run time: 0:001:12:00:00 (36 hours) Points: 1200 Results Returned: 19 Note also that for that device, all three figures the day earlier would have been zero because the work unit that completed at 03:30 would have had to have started the day before that. Anyway, this is the basis for my guess. I hope I didn't confuse you (I think I am confused now -- LOL). I am convinced that running in the "blocked" shutdown is actually reducing your contribution to the projects and that your plan to run in what you are calling "normal" mode this weekend is the best. Regards, |
||
|
|
wayndal
Cruncher Joined: Nov 23, 2005 Post Count: 17 Status: Offline Project Badges:
|
Wow... good post.
I'll keep an eye on it. If anyone wants to mess with any of this, let me know. Otherwise, I'll be on my way. Thanx for your help! We'll see how it goes from here. |
||
|
|
|