My Wingman has all client errors
25 Sep 2008 23:25:47 UTC
Topic 193951
(moderation:
Hi gary
as a mod you may be the one to ask
updated to 6
now got wingman on 11 tasks
compute error on all
comp ID1605641
administrator
cpu intel(r)
cpu's 2
op xp2
tasks 20
completed 0 as far asi can tell
all the best
garry
(gaz)
Cheers,
Gary.
My Wingman has all client errors
)
I've created a new thread for your message as it wasn't really appropriate to leave it where you posted it.
All the client errors seem to have the same cause, ie Exit Status: -226 (0xffffffffffffff1e) with the message "too many exit(0)s".
Since the computer in question is a dual and is runnining 6.x.x BOINC, I think the problem is being caused by setting the CPU usage to less than 100%. This has been showing up quite a lot recently and will continue to do so until the BOINC Devs sort out the problem.
At least he is now down to 1 task per day and the errors are being reported immediately so that the tasks can be quickly reissued. Hopefully the owner will notice the problem and will come looking for answers here.
Cheers,
Gary.
RE: Since the computer in
)
I recently brought up a new laptop on BOINC. It is a Core 2 Duo in a very, very small form factor, so the fan really snarls and the bottom gets very hot at 100%. I initially set it up with BOINC 6.whatever was on that day, but disliked a couple of things, and backed it down to 5.10.45 as are some of my others. I'm currently restricting it to one core, but plan to cut it back at least a factor of two more soon to cut fan noise and heating more.
In regard to the second point (people looking for answers), would you currently advise people who need or want to run at less than 100% to choose 5.10.45? Or are there better solutions?
RE: In regard to the second
)
I don't really know what is best to advise since I have zero personal experience of 6.x.x BOINC. I almost got tempted at one point when 6.2.14 became the recommended version but when this was reversed a short time later, I was rather pleased with my "tardiness" :-). My personal opinion is to stay clear of BOINC 6 until it has a month or three of trouble free operation. So for the people who want to use less than 100% CPU utilisation, I'd recommend staying at 5.10.45 if running at 100% is simply not an option.
Come to think of it, since I never use less than 100% myself (if it won't run satisfactorily at 100% I don't use the host), I'm not even sure the problem doesn't exist in the later 5.x.x versions as well. Maybe someone can enlighten me about that.
Cheers,
Gary.
RE: I don't really know
)
I can only second that! :-)
I'm quite sure that this problem has occurred before BOINC 6.x came along. I don't think it's a problem of the client either, but with the (BOINC code in the) project application. When I had a heat problem with my laptop (before I blew out the dust bunnies :-), I used ThreadMaster to throttle cpu usage without problems.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
My issue is similar for
)
My issue is similar for different reasons.
My old P4 dual Prestonia Xeon now has over 4 days continuous output held in pending A considerable number of these WUs have no wingmen (Server side report: unsent). The time out on these WU with no wingmen is still 2 (or more) weeks away.
I assume these WUs will need to wait to their deadline before they are resent, and then there will be a delay before they report?
Anyone else getting this level of build up on their pending credit account, for low RAC rigs? On the server side adjustment for credit, I calculate there is about 3,300 CC waiting here.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
RE: ... A considerable
)
Since you have crunched and returned tasks (which remain unsent to a wingman) they cannot timeout and you don't need to be concerned about your original deadlines. A lot of us have hosts with many completed tasks for which no wingman has yet been assigned. I've got a new fast host that has more than sixty such tasks going back several days. I've tried to explain possible reasons for this "out of sync" behaviour in a couple of recent messages such as this one and this one.
Whilst this behaviour might not be optimal for the credit conscious or those troubled by falling RACs for no apparent reason, in the end the unsent tasks will be delivered and all will be satisfactorily concluded without any intervention on the user's part. When the catch up occurs, it's likely to be rapid and will probably cause a pleasant (but temporary) boost in RAC until things settle down.
Cheers,
Gary.
Thanks Gary I was not
)
Thanks Gary
I was not worried by a loss of either RAC or credit, and I am pleased to hear I am not the only cruncher with "out of sync WUs".
All will be restored in time, then!
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
hi guys my question to
)
hi guys
my question to Gary was about two wingmen
anonymous comp ID1304376
he has 161 tasks all have returnd as client detached
administrator comp ID1605641
all his come back as compute error
i have no problems with pending got 8 at the moment
just seamed strange 161 is a lot of errors for one client
all the best
to one and all
gaz