Posts by ~cRUNch.80/dk~

1) Questions and Answers : Windows : firewalling (Message 286013)
Posted 20 Apr 2006 by Profile ~cRUNch.80/dk~
Post:
The applications do not talk onto the Internet or to project sites. boinc.exe does all that. Are you sure this is happening as you might think?

Yes, quite sure. It might be a wrongfull determination on behalf of the FW, which program is actually doing the internetting. Might it just be the fact that the project-app requests some info, or give back a trickle a la CPDN, and thereby makes boinc.exe do some trafic, wrongfully looked as as being the project-app doing it you say? This just doesn't seem too plausible, as my FW have a warning dedicated to just that kind of application "hijacking", and the warning recieved is a regular "app such and such wants to contact url/ip such and such", and the app doing the request is the project-app, not boinc.exe. That's what wonders me.
2) Questions and Answers : Windows : firewalling (Message 286011)
Posted 20 Apr 2006 by Profile ~cRUNch.80/dk~
Post:
.. Also, the BOINC app would have to be modified anytime a NEW project is added or deleted to allow for the new traffic. ......and if the project changed, then the new settings would have to be changed in BOIINC..

err... boinc is not a firewall, so why on earth would the client need any changes whatsoever if/when you add a new project, other than the change that is to setup up the client to communicate with this new project.
3) Questions and Answers : Windows : firewalling (Message 285877)
Posted 20 Apr 2006 by Profile ~cRUNch.80/dk~
Post:
Hi, I oftentimes notice that the project-app and not boinc.exe wants to do some internet-traffic. When the URL is resolved into an adress it usually concerns adress(es) that obviously relate to the projects site, but nevertheless I'm still 'worried' if this traffic should take place at all. Shouldn't it be only boinc.exe that performs all the fetching/submitting asf. not least in aspects to configuring a firewall, instead of having to setup rules for each project's exe, and reapproving the access?
4) Questions and Answers : Windows : what fw-holes to poke (Message 213873)
Posted 14 Dec 2005 by Profile ~cRUNch.80/dk~
Post:

Winblows FireWall and ZoneAlarm, you must give exception to BOINC core client command line interface and BOINC GUI

Winblows :D firewall and zonealarm has just about absolutely nothing to do with this issue!
WF is disabled, and Sygate is not Zonealarm, the question is what exact IP's need to be excluded in the xtra-level ip blocker.
Searching for an answer to this brings up a whole bunch of very different threads, so I'm mostly singing a song for this info to be included as a sort of FAQ, like mentioned earlier ... it ought to be the fourth list-item is displayed.
I've downed the range to just 128.32.18.151-128.32.18.152, which seems to be the web-relevant IP(s) and 128.32.18.173 which apparently is the IP for WU-issues.
If those 3 IP's are removed from the IP-blocker, then there appears to be no problems, but it's kinda strenuous to narrow down the filtering this way (by observing ip-blocker logs each time somethings seems to be in the way ... for EACH attached project that is ... instead of having this info readilly available as e.g. the fourth step concering how to BOINC (in this case/forum how to S@H-BOINC).
5) Questions and Answers : Windows : what fw-holes to poke (Message 190214)
Posted 18 Nov 2005 by Profile ~cRUNch.80/dk~
Post:
I'm using both a regular firewall, sygate, and "xtra-level" protection peerguardian.
Also I'm behind a router with buildin firewall, NAT'n'all that.
The problem doesn't seem to be with the router nor the sygate, but that the relevant IP's for s@h and other projects are amongst the nearly 3 bill. IP's blocked.
So far I've allowed the range 128.32.18.152-128.32.18.173, but is it the entire 128.32.18.x that ought to be allowed, and why doesn't the project (none of them) have a clear link to those firewall-info's?
Nomatter how I try to refine the search it returns a HEAP of results : the info is FAQ-relevant! =
1. Read S@H's rules and policies.
2. Get a S@H account.
3. Download and install BOINC software.
4. Allow those IP's in firewall.
6) Questions and Answers : Windows : When closing down the system. (Message 47809)
Posted 19 Nov 2004 by Profile ~cRUNch.80/dk~
Post:
OK, so I'll enterpret this statement from CPDN as a "decide for yourself"-ish 'problem'. Can ne1 then tell me how to inform windows to properly shutdown/exit boinc before the final shutting down of the system? Just to make sure. Either by macro or by inserting a statement somewhere in windows.
That would be nice :)
This issue COULD explain why so relatively many WU's turns out errors according to the results-page on my account though appearing to be correctly computed via boinc ... though I still find it hard to believe that this is for real. I mean ... if this IS an issue regarding boinc it smells a bit like .. well bad programming, no offence intended boinc-crew :-)
I would like to think that it's simply CDPN that haven't updated their pages to reflect recent changes, but perhaps someone 'inside' could confirm this?
7) Questions and Answers : Windows : When closing down the system. (Message 47270)
Posted 17 Nov 2004 by Profile ~cRUNch.80/dk~
Post:
hmm ... so i just browsed around a bit at cpdn and stumbled upon this:
"If you are running BOINC on Windows make sure you exit the program before shutting down the system. BOINC has no control over how Windows shuts things down, and you can get a corruption of the client_state.xml file which may cause the startup to fail when the system comes back up. The CLI client seems to be safe when it's run as a service, but running net stop boinc before shutting down will make sure."

As I suppose this is not just assumptions but the actual truth I'm just wondering; Is this to be changed in client 4.13+ ?

Or has it infact already been changed with cpdn just not having updated this statement?
8) Questions and Answers : Wish list : At least change the text!! (Message 47232)
Posted 17 Nov 2004 by Profile ~cRUNch.80/dk~
Post:
It's beginning to get very annoying to get the message "This feature is turned off temporarily" when attempting to view your pending credits as I've never seen this page/function running AT ALL!
Remove the "temporarily"-part at least ...
9) Questions and Answers : Wish list : Payroll Type Credit System (Message 39029)
Posted 22 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
>
> I'm not suggesting any changes to the way credits are calculated.
>
> I just want the additional info to understand the credits.
>
> : )

Yes, i know, but you have to admit that the system of creditting is unfair when regarding the possibility of getting no credits for a perfectly good WU, for whatever the reason might be, and certainly when due to errors by the project. A display of total of e.g. the claimed credit and WU CPU-time would better this though, no doubt.
10) Questions and Answers : Wish list : Payroll Type Credit System (Message 38999)
Posted 22 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
> ...
> Edit: In other other words if Boinc only gets to run .6 out of every second
> then the cpu time only increments .6 second.
> ; )
>
So yet again to put it other words: you wish for the CPU-time to be reformed into a standard, though not a standard based on actual calc's performed, but solely a 'local' standard?

I can't by any means accept that for intance a WU I spotted at LHC had been treated by 6 clients. All the WU's requested 0.18 (doh!) credit, yet only one of the 6 clients actually got any credit at all (though NO errors from either of the clients). Serious reforming need to be done in a manner, that this cannot happen. The validation-part in it self is ok, as it reduces the risk of frauding WU's in order to move up the ranks, and 'never mind' then, that you might not get credited the second the WU is returned. But the above mentioned crediting kinda ticks me off! I've already had quite a few WU's not credited , due to errors in computing, uploading, et.c. but that doesn't mean that I haven't contributed precious CPU (not to forget electricity which is an actual expence in the respective currencies for the user), but still the projects are able to say "no, that's not good enough, nothing for you on that one" and THAT is kind of a hard pill to swallow! So what that you couldn't use it? I still wish to have the computing done recognized in SOME way. I find it disrespectfull that the current method of crediting doesn't take into consideration, that we actually pay, in true currency, to support the projects. Escpecially when a perfectly good WU gets no credit due to UL-error on behalf of the projects server or whatever.
11) Questions and Answers : Wish list : Payroll Type Credit System (Message 38744)
Posted 21 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
>
> My guess would be : BoincLogX logs only results that are in the state "ready
> to report", not trickles. 500 credits is somewhere at trickle 6 or 7, one CPDN
> model has 3 phases x 25 trickles so it will grant about 75.6 x 75 credits
> before the model is done. The trickles produce a scheduler request but afaik.
> not any data to send back to the project. After finishing a phase some data
> get sent back and it adds stats about the model phase to the result page.

I'm not even sure i fully understand the concept of trickles. Or even what it actually means :) But your'e saying, that BoincLogX will only create 3 entries pr. CDPN unit? Despite their taking soo long to complete.
12) Questions and Answers : Wish list : Payroll Type Credit System (Message 38674)
Posted 21 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
> For SETI "Best score" database entries (pulse, triplet ...) in the user record
> would be great too :-)
>
Ahh the neat setispy form classic seti@home, which had great features in the field of informing what actually have come out of your WU's.
BoincLogX just doesn't give the same satisfaction, and neither does the result-log presented at your account.

Can anyone explain by the way, why BoincLogX doesn't catch the results send to CDPN? I've got more than 500 credit by CDPN since I started using BoincLogX, but BoincLogX haven't logged'em. The reason I chose to run BoincLogX, was to make my own statistic to check if BOINC distributes the resources as set up. But it's kind af rediculous to try making that statistic if it will only be more or less accurate once pr. how long it takes to complete a CPDN-WU.

And last; the CPU-time in seconds currently displayed for each result, is this simply how long time the WU have been running on my computer, or is 1 second of CPU-time a standard unit such as par example CDROM-drive speed? If solely being how long the WU have run on the computer, it's totally worthless. 1 second of CPU-time by client in the daytime isn't the same, in computingpower, as 1 second at night, since I only allow BOINC projects to utilize all it can get at night, and a maximum of 60% during the day, often less.
But i now see that's what this thread is all about.
13) Message boards : Number crunching : Stats for dummies (Message 38607)
Posted 20 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
Yes I've got the basics now, BUT:
if anyone could do a truely dummie-explanation.
The 'for dummies' part of this thread needs to be emphasized in a way that enables anyone who wish to create a stats-page, whether for just for themselves or for their team.
The optimal would probably be one of the current statspages such as boincstats.com or boinc.dk having this dummie-explanation, but if one already do, I haven't found it!
14) Questions and Answers : Wish list : BOINC should consider remaing time of WU (Message 38513)
Posted 20 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
It's happened quite a few more times than you would think. A WU have less than 10 minutes before being complete, and then BOINC change the active project according to sharefactor and the "change application every x minutes" setting. This is just folly! I'm 'only' running three projects, but still it's a matter of HOURS before boinc returns to the WU nearly finished.

When this is brought up, it's also necessary to bring up, that the seti-application ought to be better at calculating the remaining as opposed to now, where it few seconds after starting a WU thinks that about 30 minutes of computing remains. It then flipflops and start counting UPWARDS for for a good 10 minutes before starting to count down again! and not only that; the the remaining time stays at the same point for several seconds before counting one second down. What is it in the workings of SAH that makes this time-calculation so screwy and unreliable??
15) Questions and Answers : Wish list : Payroll Type Credit System (Message 38511)
Posted 20 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
I'm not all sure that even more work for the project-servers is the right way. Infact is CDPN the only server that haven't been down for extended periods yet, as far as I've noticed. But if the returned WU already contains the actual number of calculations performed by the client to complete it, or easily could be redesigned to do so, then yes .. give us the info dag nabbit :)
16) Questions and Answers : Wish list : Targeted suspension of individual projects and components (Message 38454)
Posted 20 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
> Last time that I checked it was impossible to set a resource share to 0. It
> would be nice to have a suspend after WUs completed.

Ok, I admit ... that part was an assumption. I haven't tried myself to set a share to 0. But the rest up to that would to do the trick if you detach right after upload of units. This way WU's might be downloaded before detach though, and considering that you're right: some sort of manual suspend-funtions would be the optimal, but why even join a project if you don't wish to contribute CPU to it? Or you could set the share to 1 then, THAT is possible(just checked) and the project you wish to focus on to e.g. 200. That way the cpu-time used by the project would be ridiculously low anyway and you would'nt have to detach to make sure that it doesn't consume (too much) cpu. But I guess I just don't understand the deeper reason for this wish when the basics of getting CPDN finished can be done by altering the share-factors. This requires more from the user than the suspend-function would though, no doubt, but if a relatively acceptable solution allready exists?
17) Questions and Answers : Wish list : Payroll Type Credit System (Message 38083)
Posted 19 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
I totally agree that some info of actual donated CPU-time, regardless of failed upload of result, error while computing WU et.c. would look very nice on the 'Projects'-pane in BOINC .... BUT; a simple display of how many hh:mm:ss the individual projects have been running on the computer isn't satisfying for a user like myself, who are running ThreadMaster to reduce the CPU-load at certain hours/situations. It would need to be a more precise indication of the actual CPU-work .... how many calc's, reformulated into units that doesn't end up in the insane numbers that the performed calc's would. Cobblestones.
We're propably saying the same thing, Boinc User, just in different ways ;)
18) Questions and Answers : Wish list : Targeted suspension of individual projects and components (Message 38053)
Posted 19 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
To answer based on your example:
adjusting the resource-share of S@H to zero would do just that.
And to prevent LHC from dl'ing new units while all your'e trying to, is to get rid of the LHC-queue/detach from project without your present WU's being wasted, simply "Disable BOINC Network Access" in the menu, and then when all LHC-WU's have completed go to LHC-web and set it to zero also(or detach if that's what you want). Then use the "Update"-command from the 'Projects'-pane in BIONC followed by re-allowing Boinc Network Access. This SHOULD make BOINC do the preferences-update prior to requesting new WU's thus telling BOINC that the new resource-share of LHC is zero, hence not DL new WU's.
(You say potato, I say potarto, you call'em DataUnit, I call'em WorkUnit)
19) Questions and Answers : Wish list : I wish Boinc had a setting to throttle back CPU (Message 38042)
Posted 19 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
I'm currently using ThreadMaster (hereafter TM) that is supplied in the add-on pages, which works just great. Furthermore I've begun constructing a GUI for TM as it's kinda geeky in it's setup hence not all that appealing to 'the common' man, and the GUI also implements the missing features descriped below:

It's funny though .... I wrote a mail to the seti-crew with a wish for seti@home to grow into a large scale platform for other projects as well, as I was running both D2OL and boinc, thus having to change applications manually.
About 10 minutes after sending this mail, I found the info about seti/berkeley having started up a project called BOINC, very closely matching in description what I'd just suggested in my mail ... spooky!
BUT ..: the main feature of the idea i descriped in my mail, was just excactly the greater control of the CPU-load .... and what do we get? : "Do work only between the hours of XX and XX" and "Do work while computer is in use?" ... weee! How crummy is THAT?
Due to the fact, that BOINC/windows isn't very great at yelding, leaving cycles available for the app. your'e actually working with (and BIONC minimized in the tray), the only option at the moment (if not using TM) is to suspend the activities of BOINC, while it with no problems instead could be instructed to just drop it's CPU-load to e.g. 50%

The whole idea of BIONC is to harvest the incredible ammount of cycles not being used, but the effort to really optimize this harvest just isn't made on this excact issue.
E.g.: You install BOINC and set it up to only do work when computer is idle, and only between the hours of 16:00 and 08:00 (because you're working on the computer between 08:00 and 16:00), and BOINC just sits there waiting, rolling thumbs giving the sky-look at it's programmers for not letting it use just 20% of the CPU as the hardworking man at the computer isn't using more than 30% of the CPU anyway, thereby leaving quite a lot of the CPU unused.
Which ironically is what BOINC is to prevent by using the cycles for 'the greater good'!
I've tried poking Søren (author of TM and yet another great Dane ;) ) to make a minor change to the workings of TM that would make the CPU-load issue obsolete when used with my GUI (well it allready is, and the suggested change is more of a 'cosmetic' nature). This however only applies to windows-BOINC'ers, so i would much rather see BOINC itself being able to manage the CPU-load, regardless of platform.
If BOINC changed face from being just a cradle, to being both cradle AND nurse.
But well ... MY own need for load-control is 90% satisfied by current TM, and would be satisfied 100% with the suggested cosmetic change of TM to make my GUI just perfect ;) But as TM/myGUI only applies for windows platforms, I urge the programmers to seriously take our cry-out into consideration.
(concerning the idea of BOINC controlling the load on base of the CPU-temp., Mikie Tim T is absolutely right. It's "impossible" to do that in an orderly and not continuesly man-hour-consuming fashion. Controlling the load regardless of temperatures however isn't)
20) Questions and Answers : Wish list : seti bug (Message 38022)
Posted 18 Oct 2004 by Profile ~cRUNch.80/dk~
Post:
I've noticed BOINC working on the same unit for up to 20minutes past the 100%.
It can't be related to BOINC it self, as I've seen neither LHC or CPDN do this.


Next 20


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.