Message boards :
Number crunching :
Funny thing with BOINC 4.43
Message board moderation
Author | Message |
---|---|
Send message Joined: 27 Sep 04 Posts: 11 Credit: 156,184 RAC: 0 |
I upgraded to BOINC 4.43 today and noticed that I had two Pirates@Home in the projects tab and one had the number of credits I had for LHC, but it was banging away at the server for work units and complaining that it didn't have a valid account key. I detached and reattached, and it's labled correctly now and getting the no work messages. Pam |
Send message Joined: 29 Sep 04 Posts: 196 Credit: 207,040 RAC: 0 |
Aside from problems with a beta BOINC client, recent news says there will be a pause in new WUs being issued from LHC while work gets validated and prepared amongst other things the LHC staff has up their krafty sleeves. :) So your "no work" message is normal. The only officially supported BOINC client for LHC@Home AFAIK is 4.19. A great deal of the code LHC uses behind the scenes is from the earlier releases of BOINC server software (Please correct me if I'm wrong, I like to stay informed). Also from the plethora of bug reports on the 4.3x/4.4x client development mailing list I wouldn't recommend using them .. versions have been coming out as often as every other day to fix problems like what you're experiencing. As far as your "no work error" one of the project team, Markku had this to say in another thread: Posted: 12 May 2005 11:21:27 UTC We are currently finishing "tail" of current study, that is, resubmitting work until every workunit has been processed. After that there will be a short break (week or so) and then about one million new workunits (which makes about five million results) will be submitted. Markku Degerholm LHC@home admin Hope that answers your question- without a search feature in here it seems daunting to find the answers you're looking for but they're there .. I'm still holding my breath on Berkeley updating the message board code to include search and sticky messages. :) PS: Your reset also has caused your account to have multiple hosts. You should merge them in the "Your Account" functions from the homepage. ..and unless you're on a corporate-controlled PC, you should upgrade to Windows XP Service Pack 2. :) |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
The only officially supported BOINC client for LHC@Home AFAIK is 4.19. ========== I don't know if that is true or not Travis but both the Seti Site & the LHC Site have the v4.43 listed as the Recommended Version and the 4.19 as the Older Version ... Personally I have run the v4.25 for quite some time now with no problems, I have version 4.43 installed also but I've have never been able to get any LHC WU's with it while I can get a few LHC WU's now & then with the v4.25 even though the Site says it's out of work ... :P |
Send message Joined: 27 Sep 04 Posts: 282 Credit: 1,415,417 RAC: 0 |
> The only officially supported BOINC client for LHC@Home AFAIK is 4.19. > ========== > > I don't know if that is true or not Travis but both the Seti Site & the > LHC Site have the v4.43 listed as the Recommended Version and the 4.19 as the > Older Version ... > > Personally I have run the v4.25 for quite some time now with no problems, I > have version 4.43 installed also but I've have never been able to get any LHC > WU's with it while I can get a few LHC WU's now & then with the v4.25 even > though the Site says it's out of work ... :P > http://lhcathome.cern.ch/download.php LHC lists 4.19 as the recommended version. I personally have 4.25 and 4.19 running just about fine.... Seti might be ahead of others with their recommended version, but by looking at their forums, you can find many problems with previous versions. Let's stick with the version LHC admins recommend.... :-) |
Send message Joined: 27 Jul 04 Posts: 182 Credit: 1,880 RAC: 0 |
We were thinking about updating the client version we offer. We will wait for a while and if we do not hear of any critical bugs with 4.43 we might start to offer it. Chrulle Research Assistant & Ex-LHC@home developer Niels Bohr Institute |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
> We were thinking about updating the client version we offer. We will wait for > a while and if we do not hear of any critical bugs with 4.43 we might start to > offer it. =========== I had v4.43 installed on 7 different Computers Chrulle, I never was able to get any work from LHC with that client, as soon as I switched back to v4.25 I started getting WU's again ... I don't know if it was a bug or not but I for 1 will not run the v4.43 Client again until somebody says they can actually get WU's from LHC with it ... |
Send message Joined: 2 Sep 04 Posts: 18 Credit: 72,233 RAC: 0 |
> I had v4.43 installed on 7 different Computers Chrulle, I never was able to > get any work from LHC with that client, as soon as I switched back to v4.25 I > started getting WU's again ... > > I don't know if it was a bug or not but I for 1 will not run the v4.43 Client > again until somebody says they can actually get WU's from LHC with it ... Today I got 1 WU and I have chrunched it. It was o.k. wolmic |
Send message Joined: 30 Sep 04 Posts: 112 Credit: 104,059 RAC: 0 |
> > I had v4.43 installed on 7 different Computers Chrulle, I never was able to > get any work from LHC with that client, as soon as I switched back to v4.25 I > started getting WU's again ... > > I don't know if it was a bug or not but I for 1 will not run the v4.43 Client > again until somebody says they can actually get WU's from LHC with it ... > I've been using 4.43 ever since Berkeley made it their "recommended" version, and I have received about 20 of the '1-off' work units per machine in the last few days. It works for me...now all we need are MANY WU'S to test it with.... *just to make sure it works* LOL :) |
Send message Joined: 1 Sep 04 Posts: 157 Credit: 82,604 RAC: 0 |
> I had v4.43 installed on 7 different Computers Chrulle, I never was able to > get any work from LHC with that client, as soon as I switched back to v4.25 I > started getting WU's again ... > > I don't know if it was a bug or not but I for 1 will not run the v4.43 Client > again until somebody says they can actually get WU's from LHC with it ... I'm on v4.43 since yesterday morning and Boinc went dry for the first time on Seti WU's yesterday afternoon and also for Einstein WU's today. Boinc downloaded some 5 WU's this early morning but for now no way to have it download any WU for those 2 projects even after 24/05/2005 20:29:38||May run out of work in 4.00 days; requesting more Since that message, no WU's have been downloaded. If this doesn't change after the 2 last crunched Seti WU's I think I will go back to 4.35. With that version I never had any problem at all. |
Send message Joined: 23 Oct 04 Posts: 358 Credit: 1,439,205 RAC: 0 |
> > I had v4.43 installed on 7 different Computers Chrulle, I never was able > to > > get any work from LHC with that client, as soon as I switched back to > v4.25 I > > started getting WU's again ... > > > > I don't know if it was a bug or not but I for 1 will not run the v4.43 > Client > > again until somebody says they can actually get WU's from LHC with it ... > > > I'm on v4.43 since yesterday morning and Boinc went dry for the first time on > Seti WU's yesterday afternoon and also for Einstein WU's today. Boinc > downloaded some 5 WU's this early morning but for now no way to have it > download any WU for those 2 projects even after > 24/05/2005 20:29:38||May run out of work in 4.00 days; requesting more > > Since that message, no WU's have been downloaded. If this doesn't change after > the 2 last crunched Seti WU's I think I will go back to 4.35. With that > version I never had any problem at all. > @ Thierry Van Driessche I noted the same as you did 4 days ago, what has driven me a little angry. But yesterday I installed it (CC 4.43) on my notebook. Before I deinstalled all other versions and cleaned also the folders. So I was obliged to attach at new at each project I want to participate. Since I did so it works fine. I'm attached on LHC, SAH, PAH and EAH. LHC and PAH have difficulties to send work, because their workservers are not producing any. With SAH and EAH the Client works as it should. The Client needs alot of patient!!! greetz from Switzerland littleBouncer BTW: Be carefull when you 'only' upgrade and you are attached on CPDN! After installation the client gets immediately new work from there, even when you not need them! a little hint to prevent that: After installation; the client runs the benchmarktest. At this time select CPDN on the projects tab and choose 'no more work'....;-) |
Send message Joined: 1 Sep 04 Posts: 157 Credit: 82,604 RAC: 0 |
> I noted the same as you did 4 days ago, what droves me a little angry. > But yesterday I installed it (CC 4.43) on my notebook. Before I deinstalled > all other versions and cleaned also the folders. So I was obliged to attach at > new at each project I want to participate. Since I did so it works fine. > I'm attached on LHC, SAH, PAH and EAH. LHC and PAH have difficulties to send > work, because their workservers are not producing any. With SAH and EAH the > Client works as it should. > The Client needs alot of patient!!! Well after all, 1 WU for Seti was sent back leaving only 1 WU for Seti over. At that moment, the manager downloaded 5 WU's for E@H. It seems to work but apparently the LT debt has to learn dealing with timings. That's the way I interpret it for the moment. Update: From JohnMcLeodVII Yes, the positive debt for LHC is the culprit. Yes, it should take this into account, and a fix for this has been checked in As LHC had a quite big positive LT debt on my host, this is messing up the downloads for other projects. Greetings from Belgium. |
Send message Joined: 2 Sep 04 Posts: 545 Credit: 148,912 RAC: 0 |
> As LHC had a quite big positive LT debt on my host, this is messing up the > downloads for other projects. I have a large LTD on LHC@Home and Predictor@Home and it has gone and gotten SETI@Home work at least once today ... Though it is pretty insistent that it wants LHC work ... so, it looks like it works most of the time, and John has found a problem that causes the CPDN extra work (and possibly others) as the initial time when a Work Unit is started it goes to 0 on the time when it should not ... so, if there is a hiccup in there ... you get more work than you should. So, in a dot release or so it looks like it will be pretty solid. And he is trying to get more data on the modem users issues ... |
Send message Joined: 2 Sep 04 Posts: 29 Credit: 116,231 RAC: 0 |
4.43 is the new officially recommended build ! http://boinc.berkeley.edu/download.php?dev=1 but the schedluer is still wacky it seems. i now got 2 climateprediction units guessed at ~500 hours each. and the client aint getting ANY other new work cause it thinks it has enough todo, heh. 25/05/2005 01:04:41||request_reschedule_cpus: project op 25/05/2005 01:04:41||schedule_cpus: must schedule is all it always says... |
Send message Joined: 29 Sep 04 Posts: 196 Credit: 207,040 RAC: 0 |
> 4.43 is the new officially recommended build ! ...from Berkeley. Check each project for their recommendations and/or read the message boards. CPDN's Windows board has several messages with 4.43 problems. Just in case anyone's wondering why I have negative issues with the new builds- when the 4.6 branch became the 4.2 branch, I was running one of those 4.2x beta builds which quickly made my system freeze. Upon restarting, I received a message that my BIOS was corrupted and had to be reflashed before it would continue. Fortunately Biostar has a Bios ROM onboard for situations like this (and the average user wouldn't know what to do for that matter). Point is BOINC could have cost me a motherboard. I have **zero** issues with 4.19 aside the benchmark optmization on the Windows platform. 4.25 is installed on my server and it works well provided it wasn't upgraded from 4.19. The boinc_dev mailing list details problems with even the most recent builds (give or take it's normal for beta software to have issues). It seems to me (and I'm honestly not trying to be jaded here) but when a "recommended" version comes out lately it feels like it's a beta client that was renamed to a recommended client, and issues continue as this thread and other threads like it in other projects indicate. Beta testing has its place, but IMHO, should that take place on a non-production server, such as LHC@Home Alpha? AFAIK all BOINC client beta tests occur on production servers with minimal interaction with the programmers of the projects the clients attach to. I'd worry the beta clients could potentially do something stupid to the work they're sent and return "bad" data to the project. /end rant |
Send message Joined: 2 Sep 04 Posts: 378 Credit: 10,765 RAC: 0 |
Myself, I'm just not a fan of the 'boinc manager' interface. I like the 4.19 version. |
Send message Joined: 24 Oct 04 Posts: 18 Credit: 1,485 RAC: 0 |
I upgraded just a while ago and apart from getting more unwanted work from CPDN, I had this message 5/27/2005 5:32:22 PM|LHC@home|Deferring communication with project for 2 days, 14 hours, 16 minutes, and 42 seconds 5/27/2005 5:35:32 PM||Nearly overcommitted. 5/27/2005 5:35:32 PM||New work fetch policy: no work fetch allowed. What does "Nearly overcommitted" mean? Thanks in advance! WinXP Home, P4 3.2GHz HT, 512Mb RAM./Win2K, P4 1.8GHz, 512MB RAM Join World Wide S.E.T.I. |
Send message Joined: 2 Sep 04 Posts: 352 Credit: 1,393,150 RAC: 0 |
To me it seems the whole BOINC Manager thingy is moving more & more toward where you are being forced to run what they want you to run & not what you want to run. The v4.43 is a prime example of this movement. I've yet been able to Download what I want, only what the Manager whats to give me no matter what I set my Preferences to ... If it gets to the point I'm only running what the Projects as a whole wants to give me & it's not what I want to run I'll give it serious consideration of just not running them at all ... I look at it this way, it's costing me between $50 & $70 a month in Electrical Bills to run the Projects & I will either run what I want to run or nothing at all ... |
Send message Joined: 24 Oct 04 Posts: 18 Credit: 1,485 RAC: 0 |
Another funny thing... 5/27/2005 7:32:24 PM|LHC@home|Deferring communication with project for 2 days, 12 hours, 16 minutes, and 41 seconds 5/27/2005 8:24:06 PM||schedule_cpus: time 3600.000000 what does this "time 3600.000000" mean? WinXP Home, P4 3.2GHz HT, 512Mb RAM./Win2K, P4 1.8GHz, 512MB RAM Join World Wide S.E.T.I. |
Send message Joined: 23 Oct 04 Posts: 358 Credit: 1,439,205 RAC: 0 |
> Another funny thing... > > 5/27/2005 7:32:24 PM|LHC@home|Deferring communication with project for 2 days, > 12 hours, 16 minutes, and 41 seconds > 5/27/2005 8:24:06 PM||schedule_cpus: time 3600.000000 > > what does this "time 3600.000000" mean? ----- "time 3600.000000" = 1 hour or 3600 sec. ...;-) greetz littleBouncer |
Send message Joined: 17 Sep 04 Posts: 190 Credit: 649,637 RAC: 0 |
@PB right now using the dev version 4.44. this one is "better" immo. The 4.44 IS writing proper the values of dept and long term dept into the client state. For me, this version looks better than prior versions. But basically as "owner" of the bills, your are right. The 4.19 was immo more user friendly, depending the point of view. Other client releases, I do like more than the newer ones. Also there is a lack of information and the client still has some (heavy) bugs. Bugs which are reported and it will take a while until everything will run smooth. Heja! My point of view is as simple as I am, just let me crunch! Just let get the clients the work, so far nearly never had any problems with dealined works. What is for sure, the newer version are getting less work than the prior versions. ________________________________________________________________ > Another funny thing... > > 5/27/2005 7:32:24 PM|LHC@home|Deferring communication with project for 2 days, > 12 hours, 16 minutes, and 41 seconds > 5/27/2005 8:24:06 PM||schedule_cpus: time 3600.000000 > > what does this "time 3600.000000" mean? that's 3600 seconds --> one hour. ________________________________________________________________ >Nearly "overcommitted" well, my brain is going to be overcommitted due to much complicated clients ;) Nearly overcommitted and overcommitted are new "status/feedback" messages. In the meaning, if the clients has enough work from one project, depending the preferences settings, when it is going to be "full" of work or nearly full of work, then those messages appears. But attention, for example, if the Ressource Shares from a project is very low and the client is attached to several projects, it happened (more than one time) to have a clients (4.43) to be overcommitted with just one(1) work unit in the queue. For predictor was using a resulting share of less than 1%. This worked fine (with 4.19) for getting just 1-2 WUs from there. With the greater than 4.3x clients, this I can forget. I had to adjust all the ressource settings. |
©2024 CERN