Einstein says it is reluctant to give work units, but gives them
17 Apr 2005 12:12:26 UTC
Topic 188973
(moderation:
First of all: This is not a huge big deal. At the end, Einstein gives me work units, which is the point. Along the way, Einstein gives some improvable error messages.
Here's a set of error messages:
04/16/2005 2:41:14 PM||May run out of work in 0.01 days; requesting more
04/16/2005 2:41:14 PM|Einstein@Home|Requesting 863.52 seconds of work
04/16/2005 2:41:14 PM|Einstein@Home|Sending request to scheduler: http://einstein.phys.uwm.edu/EinsteinAtHome_cgi/cgi
04/16/2005 2:41:15 PM|Einstein@Home|Scheduler RPC to http://einstein.phys.uwm.edu/EinsteinAtHome_cgi/cgi succeeded
04/16/2005 2:41:15 PM|Einstein@Home|Message from server: No work sent
04/16/2005 2:41:15 PM|Einstein@Home|Message from server: (won't finish in time) Computer on 44.2% of time, BOINC on 61.4% of that, this project gets 45.0% of that
04/16/2005 2:41:15 PM|Einstein@Home|No work from project
04/16/2005 2:41:15 PM|Einstein@Home|Deferring communication with project for 1 hours, 0 minutes, and 0 seconds
04/16/2005 2:41:16 PM|Einstein@Home|Computation for result H1_1194.4__1194.8_0.1_T05_Run2 finished
04/16/2005 2:41:16 PM|SETI@home|Restarting result 20ja05ab.15273.26914.29822.82_2 using setiathome version 4.09
04/16/2005 2:41:17 PM|Einstein@Home|Started upload of H1_1194.4__1194.8_0.1_T05_Run2_2_0
04/16/2005 2:41:19 PM|Einstein@Home|Finished upload of H1_1194.4__1194.8_0.1_T05_Run2_2_0
04/16/2005 2:41:19 PM|Einstein@Home|Throughput 55123 bytes/sec
04/16/2005 3:41:16 PM|Einstein@Home|Sending request to scheduler: http://einstein.phys.uwm.edu/EinsteinAtHome_cgi/cgi
04/16/2005 3:41:18 PM|Einstein@Home|Scheduler RPC to http://einstein.phys.uwm.edu/EinsteinAtHome_cgi/cgi succeeded
04/16/2005 3:41:18 PM|SETI@home|Pausing result 20ja05ab.15273.26914.29822.82_2 (removed from memory)
04/16/2005 3:41:18 PM|Einstein@Home|Starting result H1_1194.4__1194.7_0.1_T17_Run2_0 using einstein version 4.79
In that sequence, we see that Einstein recognizes it needs another work unit because it is almost done with the one it is working.
The server tells the client, unexplainably, "Computer on 44.2% of time, BOINC on 61.4% of that, this project gets 45.0% of that". The 44% is probably right. The 61% is unexpected; BOINC is on as a service anytime the computer is on. Einstein is indeed set for 45% of BOINC. Another project is both set for 10% and suspended, so the two projects set for 45% in fact get 50% each.
One couldn't be surprised, having seen this message, if the server declined sending a work unit. Consistent with that, there's no message indicating a download. On the other hand, in the same second as the successful second contact with the server, work starts on a new unit. (Hmmm . . . where did that unit come from? he he)
SETI responds to downloads this way.
04/16/2005 1:30:16 AM|SETI@home|Sending request to scheduler: http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi
04/16/2005 1:30:18 AM|SETI@home|Scheduler RPC to http://setiboinc.ssl.berkeley.edu/sah_cgi/cgi succeeded
04/16/2005 1:30:19 AM|SETI@home|Started download of 20ja05ab.15273.26914.29822.82
04/16/2005 1:30:25 AM|SETI@home|Finished download of 20ja05ab.15273.26914.29822.82
04/16/2005 1:30:25 AM|SETI@home|Throughput 78236 bytes/sec
Again, this is not a big deal. Thanks for letting me participate in your science.
Einstein says it is reluctant to give work units, but gives them
)
Often downloading work with this project is mearly getting a set of parameters during a RPC, the big download only happens every few weeks. That is why you don't see anything download but you have more work.
The reason it refuses at first is because the calculations indicate your computer will not finish. Once there is no E@H work on your computer you will be sent another workunit even if the calculation predicts you will not be able to finish it.
The on_frac and active_frac will get more accurate with time. The way they are calculated was changed between versions 4.1x and 4.2x so that may be part of the problem. There have also been enough posts from people with strange numbers to indicate there is a possiblity of them getting corrupted as well.
BOINC WIKI
BOINCing since 2002/12/8
> Often downloading work with
)
> Often downloading work with this project is mearly getting a set of parameters
> during a RPC, the big download only happens every few weeks. That is why you
> don't see anything download but you have more work.
Hello John,
according to the news,
April 7, 2005
During the next few days we will finish the first round of S3 analysis and begin a second round. As we finish up the first round of S3 analysis, users may find that they are getting a new data file with each workunit; users with modem connections may want to suspend Einstein@Home for a few days to avoid multiple long downloads. Note that there may also be some instability in a few days as we shift to new workunits,
I saw since yesterday I get for each new WU a new file of some 14.xxMB. If things have changed here it will indeed be hard for regular classic modem users to continue for the moment.
Greetings from Belgium
Thierry
I'm having a similar problem
)
I'm having a similar problem but the server is absolutely refusing to give me work units. I have a work unit about 2 1/2 hours from completion. I also have one seti work unit running and one ready to run (true dual processor P3 machine). I have my connection interval set to 3 days. It takes my computer just over 24 hours to complete a work unit.
Yesterday it ran three seti units and one einstein unit.
Most recent message as follows (common)
4/28/2005 10:04:00 PM|Einstein@Home|Message from server: No work sent
4/28/2005 10:04:00 PM|Einstein@Home|Message from server: (won't finish in time) Computer on 33.8% of time, BOINC on 32.0% of that, this project gets 50.0% of that
4/28/2005 10:04:00 PM|Einstein@Home|No work from project
4/28/2005 10:04:00 PM|Einstein@Home|Deferring communication with project for 4 hours, 52 minutes, and 8 seconds
My current work unit will be completed long before the next scheduled communication.
There's a problem with computer on time. Except for hardware related down time my computer is usually running non-stop for anywhere from several days to two weeks between restarts. It's certainly not down 66.2% of the time. I had a problem with my computer a few weeks ago and it was inoperative for a couple of weeks. This accounts for part of the down time, but not all of it.
I also had a problem with BOINC version 4.19 causing my TV tuner to go unresponsive so I would sometimes disable BOINC and have it run only when I wasn't using the computer. I recently updated to BOINC 4.32 and noticed it doesn't conflict with my tuner as bad and have had it running continuously for the last few days.
How does einstein@home figure that it only gets half of the BOINC running time? I have dual P3 700MHz processors and run seti@home and einstein@home work units at the same time.
Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)
S@H is getting the other 50%.
)
S@H is getting the other 50%. You will get a new WU when this one is completely done.
BOINC WIKI
> S@H is getting the other
)
> S@H is getting the other 50%. You will get a new WU when this one is
> completely done.
>
I don't think you understand my point. My computer has two distinct separate processors. The hosted applications each use one processor. Each project is getting complete use of one of my processors. e@h operation is not affected in any way by the fact that s@h is running too.
To say that it gets 50% of the BOINC running time is like saying I can't hold something in my left hand because I'm already holding something in my right hand.
I'd like to have a work unit cached and ready to run upon the completion of the running work unit. Or ready to run when I am unable to get work from another project. I don't like the idea of having to wait until 1am, then 1:30am, then 2am ... to get a new work unit. It would be nice to be able to connect once, download enough work to keep my system occupied until a reasonable time the next day, and let it go.
Nothing travels faster than the speed of light with the possible exception of bad news, which obeys its own special laws.
Douglas Adams (1952 - 2001)
From your previous post about
)
From your previous post about downtimes the time stats seem reasonable. This is mostly due to the time with the TV tuner problem and running 4.19. The time when the entire system was down should be mostly cleared out now, but still affecting this some. In a couple of weeks the stats should more accurately reflect your current computer usage and this problem should go away on it's own.
The fact that you have 2 CPUs should be figured into the request.
BOINC WIKI
BOINCing since 2002/12/8