Posts by Lee Carre

21) Message boards : Number crunching : problem downloading a WU - HTTP 416 (Message 13297)
Posted 9 Apr 2006 by Lee Carre
Post:
That's a BOINC problem, rather than one specific to Rosetta.
thought as much, just that i've never had this happened with any project before, so thought it might have been a configuration issue specific to rosetta (telling boinc to request data the wrong way prehaps, from what i saw in the capture, boinc was requesting the data as bytes, but i think the server wanted to deliver the files in gzip format, i could be wrong though)

By any chance, are you behind an HTTP proxy?
as stated yes, i normally use an HTTP proxy, but i did try without (i only use a proxy for performance reasons, but have a direct connection to the net available if needed) however a direct connection returned exactly the same response

Go into the Boinc/projects/boinc.bakerlab.org_rosetta directory, look for the file in question, and if it's there, delete it. Then try the download again, and see what happens.
surprisingly this worked, the files were present, i stopped the client and appended .bak to the end of each (just incase i needed them back later) restarted boinc and it happily started downloading the files (with the option to use a proxy enabled as well), quite odd

i guess boinc was rather confused about something if a file/disk issue causes it to have networking trouble preventing it from downloading a file lol
i fail to see how removing the files caused to to request the file with the correct headers, but then i'm not a programmer

but thanks anyway, the WU is in my queue ready to crunch now, i'll post back if i have any further issues with this


on a slightly seperate issue, as you seem to know about the innards of BOINC
i've got a working socks proxy on my network, other apps are able to use it, however when i tried to use it with boinc (after checking that i was using the correct settings) it didn't want to co-operate, is this also a known issue (prehaps a socks version issue, as v4 and v5 are quite different)
22) Message boards : Number crunching : Miscellaneous Work Unit Errors (Message 13191)
Posted 7 Apr 2006 by Lee Carre
Post:
not a processing error, but a download error, details in the problem downloading a WU - HTTP 416 thread
23) Message boards : Number crunching : How much of the memory req. is for the graphics? (Message 13166)
Posted 7 Apr 2006 by Lee Carre
Post:
compared to the amount of data needed for the structure i imagine a small table of values isn't much

as for memory usage, as far as i know the graphics are rendered in real-time, so i don't think there's much memory used for them (except graphics memory on your GPU)
24) Message boards : Number crunching : problem downloading a WU - HTTP 416 (Message 13165)
Posted 7 Apr 2006 by Lee Carre
Post:
boinc informs me that it's having trouble downloading a WU (7447_largescale_large_fullatom_relax_dec7447_1_01_1.pdb_422_12_0)

in the message log boinc just says "error 500", now we all know it just says 500 for a veriety of things, and that's a seperate issue, however because i've never had this problem with rosetta before i thought i'd look into it, and did a capture with Ethereal, which revealed that the server is returning "HTTP 416 Requested Range Not Satisfiable"

i'm using a standard client, and i can email the ethereal capture to anyone who's interested in looking, incase they might be able to find out why the server is returning a 416 (i guess there's something wrong with the request BOINC is sending)

i just thought i'd bring this to the attention of the admins especially, however i'm not sure if this is a rosetta problem or a boinc problem

like i said, anyone who's interested in the capture (for 2 files with the same error) please send an email to leec100@hotmail.com with a subject of "BOINC HTTP 416"
anything else will be filtered as spam


if an admin/moderator would like to make comment if this should go in the Miscellaneous Work Unit Errors thread then i'd be happy to repost there

feedback from anyone else is welcome

i've not changed anything except for increasing my "connect to network" setting recently, i normally use a proxy server, but i tried without and just got the same error (boinc saying 500, ethereal saying 416 for both files)

incase it helps/matters, the URLs for the files are:
http://boinc.bakerlab.org/rosetta/download/114/boinc_bb_3mer_paths.txt
http://boinc.bakerlab.org/rosetta/download/6d/dec7447_1_01_1.loopfile.gz

oh, i'm able to download other rosetta WUs just fine, and have no problems with other projects
25) Message boards : Number crunching : BOINC service & P4's (Message 12800)
Posted 29 Mar 2006 by Lee Carre
Post:
for both of you, try running the service under the system account

instructions can be found in the Enabling The Graphics Capability of the BOINC Client Software for a Service Type Installation guide
26) Message boards : Number crunching : Why no Interchange of Info among BOINC Projects? (Message 12799)
Posted 29 Mar 2006 by Lee Carre
Post:
In that case why not set up a formal method of info exchange between the Tech Depts of all the projects, even if it is only a regular 'round robin' e-mail with the 'ramblings' of various technical experts of the problems noted/solved on their various home projects?
as john implied, the berkeley email lists are sort of how infomation is exchanged at present, it's generally quite good and things happen quickly

a problem with the scheduler sending the same result to multiple hosts was identified at einstein, and within a few days it was resolved and a new scheduler implemented and the code added to the CVS
27) Message boards : Number crunching : BOINC service & P4's (Message 12030)
Posted 15 Mar 2006 by Lee Carre
Post:
as a quick-fix, you could set the service properties to restart the service when it fails, i assume you know where to look, if not let me know and i've write a step-by-step for you

but obviosuly this is just a temp fix, and you should solve the actual cause of the problem
28) Message boards : Number crunching : Optimized Client? (Message 12007)
Posted 14 Mar 2006 by Lee Carre
Post:
If you're looking to crunch more WU's, an optimized app won't work since Rosetta WU's are now CPU run-time dependent.
ahem, yes it will work, it'll just get more done in the same amount of time, which is exactly what rosetta needs, therefore fewer WUs need to be sent out

nevermind the fact that the science gets one faster, leading to usable results being available sooner
29) Message boards : Number crunching : BOINC service & P4's (Message 11994)
Posted 14 Mar 2006 by Lee Carre
Post:
It will run for several days, then mysteriously stop.
the reason for it stopping/crashing should be available in the windows event log

does anything else happen/change around/after the time the service stops?

is there any software that might be affecting BOINC, such as the type that restricts access or activity (some isn't that well writen)

how do you monitor boinc on your farm? i assume you use boinview, what's the refresh interval set to?

does the service stop at a particular time of day?
maybe a virus scan or other regular task is causing problems, i know that certain virus scanning software causes problems with CPDN, maybe something similar is happening here

is it just the BOINC service that stops, or do other services/processes stop as well?

what are the hosts used for during the day? (or whenever they're being used by people and not just crunching)

is there anything you've tried to discover the cause/solve the issue already?
i assume you've done basic things like runing a disk check, and run various stress tests, especially memtest
30) Message boards : Number crunching : Miscellaneous Work Unit Errors (Message 11992)
Posted 13 Mar 2006 by Lee Carre
Post:
result FA_RLXac_hom029_1acf__359_123_0 produced: exit code -1073741811 (0xc000000d) after 1 minute and 2 seconds

i have leave apps in memory set to yes, running on winXP with an Intel CPU
31) Message boards : Number crunching : Rosetta optimization (SSE2...) ? (Message 11985)
Posted 13 Mar 2006 by Lee Carre
Post:
i use BoincView which can tell me the FLOPS (also called other things like FIOps)

rosetta wasn't bad before the recently released apps, it was just above par with other apps for sheer processing speed

but with v4.82 well, the speed of some units is pretty damn fast, so i think the code is pretty efficient already

currently rosetta is the fastest app i've seen (i crunch for most but not all projects)
32) Message boards : Number crunching : Server Time wrong! (Message 11984)
Posted 13 Mar 2006 by Lee Carre
Post:
if you use a HTTP viewer to see the response headers, the "Date:" reported is pretty accurate
and that value is direct from the system clock, i imagine the server has a syncroniser set up, as correct time is important in a network environment

so this is something software related, unless the time has been corrected recently

[edit]looking at the timestamp on my post it seems correct as well[/edit]
33) Questions and Answers : Web site : I do not understand what to do with the jobs as I am very new here (Message 11846)
Posted 10 Mar 2006 by Lee Carre
Post:
all you need to do is make sure boinc has enough time to work, so that it can finish the job, but your computer itself will process the work
34) Questions and Answers : Windows : Computation erros (Message 11840)
Posted 10 Mar 2006 by Lee Carre
Post:
@lee-

TY for your responce.

"Leave applications in memory while preempted?",was set on NO,changed it to yes.

as far as tests, i went to PCPITSTOP.com & did their FULL TEST & everything including my memory passed with flying colours. no idea how thorough their tests are.

AFTER THAT i updated my ATI display drivers.

now it SEEMS to be running ok since then,on the BBC/CCE forum someone looked at the error # or whatever & said they thought it might have something to do with the display drivers & other users who had went & updated their drivers seemed to cure the problum,well it SEEMS to have worked for me also. time will tell,so far so good.

& BTW, i always suspend BOTH network activity & boinc whenever playing my online games.

if it's working now then good :)
i'm not sure about their tests either, but if you have any further trouble run thru UKnicks list of tests, they're pretty through, and will find any hardware issues, they're the tests i recommend to everyone

the issue with graphics drivers is interesting, i never would have thought graphics drivers would have caused rosetta to crash, but that's computers for you lol

anyway, happy crunching
35) Message boards : Number crunching : Is something markedly different with the newer WUs? (Message 11814)
Posted 9 Mar 2006 by Lee Carre
Post:
If the software is changed or the methodology has changed and it is up to me to make a modification, an email to me would have alerted me. If one spammer working from his living room can send out hundreds of thousands of emails several times a day, one bulk emailing informing users of the change would not have been hard to do.

not everyone wants an email, if you want to stay up to date with what's going on, get an RSS reader and subscribe to the various project news feeds out there
36) Message boards : Number crunching : Running Boinc as a Service (Message 11809)
Posted 9 Mar 2006 by Lee Carre
Post:
37) Message boards : Number crunching : Running Boinc as a Service (Message 11808)
Posted 9 Mar 2006 by Lee Carre
Post:
When I rebooted my system at home today, I had to type in my password (you have to add a password to an account with admin rights so Boinc can use that account and password to run.) and then the boinc manager pops up when my desktop appeared.

There's usually a number 8 in my taskbar (the boinc icon) - although it was missing before I rebooted. (Boinc and Rosetta were running and keeping the cpu at 100%, though.)

the icon is the manager, it's not needed when running as a service, so don't bother unless you need to manage the client, it just take processing cycles otherwise
38) Message boards : Number crunching : Running Boinc as a Service (Message 11807)
Posted 9 Mar 2006 by Lee Carre
Post:
You have to start it manually, but it works really fine. I have ran Boinc as a service on another computer and haven't faced problems. (Except the graphics don't work and that is clearly said in installation instructions.)

well no, it depends if it's set to start automatically
is it in your startup folder on the start menu? if so then it'll auto-start (which is pointless for a service install) if you go to the properties of the link (in the startup folder) then select "hide", it'll no longer auto-start (but the service will run regardless, the client and the manager are now totally seperate and boinc is started when your system boots, not when the manager starts)

as for graphics, there's an easy fix, if you know about services, load the services.msc panel, find the boinc service and go to properties, then the "log on" (or "log on as", i can't remember) tab and choose the option to run under the "system user" and tick the box below to allow it to interact with the desktop, then click ok and exit the services panel

more detailed instructions can be found in the Wiki: Enabling The Graphics Capability of the BOINC Client Software for a Service Type Installation
39) Questions and Answers : Windows : Computation erros (Message 11806)
Posted 9 Mar 2006 by Lee Carre
Post:
Iv had been running boinc with just 'seti@home' for a few months with no problums what so ever & have just resently added both the 'BBC/CCE' & 'Rosetta' projects for a total of three projects.

now with both the BBC/CCE & rosetta projects im having thease 'Computation Error's' every day,real bad with the rosetta project.

WHATS UP WIT DAT?!

what is a "Computation Error" & what is causing this pls & ty? :-)
_______________________________________________________________________________

"have you checked that your computer is stable?" = NO.

im using a gaming computer that plays HUGELY graphically intensive 'massive muti-player online games' with no problum, if my puter cant handle rosetta?! then i give up! LOL!.

in your general preferences, make sure you've set "Leave applications in memory while preempted?" to "yes"

also if the problems only occour when you're gaming, then set boinc to "run based on preferences" or suspend it if you're using a game controller (something other than the mouse/keyboad)

there is a vast difference between gaming and the kind of processing rosetta does, gaming is relatively easy, and it doesn't matter too much if errors occour, rosetta on the other hand is very sensitive to errors, and uses parts of your CPU that a game will not
so a game may work perfectly, but it doesn't mean your system is stable enough to run rosetta

run thru the list of items on the stability check list as this will identify the most likely source of the problem

for questions such as "what is a computation error?" please read the BOINC Wiki, but basically it means that something went wrong with the calculations, and the most likely cause is a hardware related issue (cooling or overclocking ususally)

please suspend any projects that are resulting in lots of errors until the problem has been resolved, because it just slows things down, and you might as well crunch for a project that gives good results

actually thinking about it, do you return good results to seti?
go to your (SETI) account, click the "my results" link and check that you're being granted credit, and you're returning valid results
40) Message boards : Number crunching : Miscellaneous Work Unit Errors (Message 11805)
Posted 9 Mar 2006 by Lee Carre
Post:
3 more:

HOMSdt_homDB009_1dtj__352_458_1 - exited after 124.2 seconds
HOMSdt_homDB003_1dtj__352_40_2 - exited after 129.9 seconds
HOMSdt_homDB027_1dtj__352_1364_1 - exited after 47.58 seconds


all returned the following error:
Incorrect function. (0x1) - exit code 1 (0x1)
my hosts are pretty stable, i have very few errors normally, and i'm using the latest recommended version of BOINC (v5.2.13)

the first 2 occoured on the same host, and the 3rd on a different host

it seems that those units have filed on all systems they've been sent to (see the WU page for each) which wouldn't surprise me if "incorrect function" means that there was something wrong with the input

[edit]oops, just seen the previous post, sorry, nevermind lol[/edit]


Previous 20 · Next 20



©2024 University of Washington
https://www.bakerlab.org