Putting up a few e0n workunits...

[SB2]

....started a few machines running e0n, more to come as time allows. ;)

[NeoGen]

That is always good news! The more, the better! ;)

[AMDave]

WU ! WU ! chugga chugga WU ! WU !

;) I know. I'm lame :lol:

Great to have you back SB2 :!:

[SB2]

Thanks NeoGen and AMDave, great to be amounst friends once again.








Time to crank it up a notch too! :cool:

[SB2]

Had enough of this e0n stuff. It would not be bad if one could get a connection to thier server when needed. The workunits are short, 3 mins or so, but I seem to be unable to connect right way and the client(s) sleep anywhere from 1 to 5 mins while waiting on a connection. So I'm wasting 50% or more of my time. :-(

[NVM]

i've had similar problems, it sux. waiting for the WU takes longer than crunching the WU.... :-(

[vaughan]

Try running something else alongside it to take up the slack. I'm doing DHEP.

[AMDave]

I wonder if their server NIC is reaching the I/O limit?

I barely got a WU this evening.

[Brucifer]

:( Well I'm sure not having a good day of it with eon. Seems as though I spend more time waiting for work units than actually working on them. So I think I'll give it up until it straightens out again and switch over to RC5-72 and crunch that for a while. Will leave a couple systems on eon though.

[AMDave]

Ya. I left a post on the eOn forum.
It seems to be picking up again this morning (for me anyway) but yesterday was very slow.

[Brucifer]

Just having a snot of a time today with eon, and not being able to connect to the server for new work units. Seems like every time the demand gets really heavy that the old delay issue starts cropping up again. I see today that one of the KWSN is really wracking up the points, as is a Team Ninja person. From the number of paoints they are adding, must be in the neighborhood of 40 added systems. Something is still wrong in their server side of the project. Just isn't worth wasting all the cpu time.
:(

So I've pulled all my systems off eon, and have put some on rc5 and TSC., and will see how things go.

[Empty_5oul]

we can expect a boost in these other two then.
TSC will be very helpful as there are a few nearby teams.

[JackindBox]

incase your wondering about the boost in EON


Apr 13,2005: Windows Client Speedup.
Features: Windows clients should now download a new library AUTOMATICALLY and work three times as fast compared with previous library.

which is great 'cept i don't think their servers can handle it...infact as soon as kwsn stomps us,im switching till they get em fixed

[Brucifer]

Except the boost in the windows client still doesn't touch the linux client. :)

Graeme said he finally saw what everyone has been talking about so now they are working on that. But then, they have been working on this issue for a long long time. :)

[Empty_5oul]

whats the SSS in the above sig ???

[JackindBox]

it's a intra team competition Empty_5oul
i dunno the exact formula but it takes all the DC projects we do,factors in the current position i hold (in Team) and gives a score based on that rank..times it by,,,,,,,,arg i dunno how it works :lol:

i do know im in 13 place against my team mates with a 52. whatever average score.. Clear as Mud,huh? :roll:

[Empty_5oul]

ah, k.

[Brucifer]

it's a intra team competition Empty_5oul
i dunno the exact formula but it takes all the DC projects we do,factors in the current position i hold (in Team) and gives a score based on that rank..times it by,,,,,,,,arg i dunno how it works :lol:

i do know im in 13 place against my team mates with a 52. whatever average score.. Clear as Mud,huh? :roll:


that is a pretty clear description!!! :roll:

:D

[Brucifer]

I've had some eon boxes running for a while now watching how the loading is going. Seems to be doing okay so far, so I added some more to it. Graeme says he has increased the size of the work units to cut down on communication with the server. So that will cut down on the daily scores to some degree I guess. But it keeps everyone in the same boat at any rate. Sure would be nice to see a *real* fix to the issue though. :cry:

But meanwhile will continue to ride the broken horse as long as the systems keep busy.

[Brucifer]

Well, it looks like Graeme really increased the size of the work units. I am putting out less than 2/3rd's of the results that I was before the resizing of the work units. So them points aren't going to be racking up so fast for a lotta folks now.

[Ototero]

At least you're getting results. I tried eon on 2 machines, ran for days and days. But no results, so I gave up.

I was trying to get a score on every project.

[AMDave]

At least we are getting WUs now.
You must have had a better rate of connection to the server than me.
My output rate has increased according to Bok's stats.
That is, I am now doing the around same output rate as before per stats update, but now in only 30 minutes instead of 60.
Until today my Linux clients were skipping a lot just waiting for the server. (Although about 1 hour ago 1 box just showed about 40 x 60 second skips in a row waiting for a connection but has resumed now)
For the timebeing this is good for me. But if it turns out to be poor for a majority, he will have to change it back as it will drop the total production rate for the project.

[AMDave]

:cry: should have knocked on wood.
It all turned to runny brown stuff.
even Bok's stats have not updated for 2 hours
I posted graeme in the forum.

[Brucifer]

Hmmm..... indeed a puzzlement. Mine are running okay. On the linux systems, make sure you have a library file named libaloptc.so in your eon_client directory. If you don't, then your system isn't going to register results correctly or talk to their server worth a hoot. I found 7 of mine bad late yesterday afternoon. So I just re-installed on all my systems. I've had much much better luck in communicating with their server. However the WU's are taking longer to process now as I mentioned previously.

[vaughan]

72CJ5 yes I've got one of them in my eon_client directory but I have heaps of ZZZ - Sleep for 60 seconds
happening now.

[Brucifer]

72CJ5 yes I've got one of them in my eon_client directory but I have heaps of ZZZ - Sleep for 60 seconds
happening now.


Wow........ gotta have something else going on too..... I've got 18 systems that i've been watching real close for two hours now, and no delays on any of them. For once. However I was getting tons of the stuff yesterday when others were saying that they were okay.

Did you try deleting the files in the directory and re-installing the client? If not, then do that as a try. Reason I mention it is that I had a box that was rolling along okay from looking at the screen, but it was not registering any points on the eon web site. It was the only box I had on eon at the time. Really blew my mind as absolutely nothing was registering. So I deleted all the files in the directory, downloaded the eon_client.tar.gz file, blew it apart in the directory, re-registered my email addy, and just like magic points started accumulating from that system.

That was the incident that started me in on re-installing. Well then in that process I noticed on some of the others that I had the wrong library file. So there are still multiple issues involved in the connection problems I believe.

[AMDave]

The I/O data is being collected., so hopefully graeme will be able to advise better once they have checked the server logs and have accumulated enough data after the change.
I'm keeping my clients online and I'll check back in 24 hours.

[Brucifer]

Graeme just posted on the EON web site hosted by FDC that the eon server is down until Monday. :-(

[SB2]

:-( Just when I was thinking of giving it another go........ :cry:

[Brucifer]

Graeme and his team have been working hard trying to find out where the bottleneck is. It's really been a rough exercise for them. While they didn't seem to have this issue when the project was at UW, they had other issues. Well they managed to get rid of other issues, but this connectivity thing has really thrown them for a loop. To their credit though, they have been very attentive to the comments and suggestions of the crunching participants which is a major plus. I think he finally decided to just pull the plug for the weekend to give them time to think things through and try to get another perspective on this without having to continually babysit the server and not be able to devote their attention to what is causing the problem.

Come Monday morning though, I'm back at it!! :D

[Brucifer]

well I guess I'm not... Seems as though the stats are really really hosed up today. Points missing, etc. sigh :(

[Brucifer]

Seems like eon is back into the zzz world again. While I like the project and what Graeme is trying to accomplish, it really gets old having systems wasting a lot of time. for a while I thought they had it running decent, but since they went and "fixed" it, it's been much worse than before. :(

[vaughan]

I agree 72CJ5 I too thought they were improving but the last few days have been bad. I run a second project with it to pickup the spare cycles when eOn goes to sleep ... ZZZs. One worth trying is D2OL if you like "Medical" projects. Alternatively, there is Seventeen or Bust for the Mathematically inclined. For the masochist there is a huge range of BOINC applications to run. I find that some of them run at lower priority than others. I don't know how to adjust this. Einstein AT home, for example, likes to take the CPU cycle away from eOn but Sztaki lets eOn take precedence.

NeoGen as our resident BOINC guru (or anyone who knows) is there a way to alter the CPU priority within a BOINC application?

[Brucifer]

I just went back to rc5 for a while, and loaded the systems up with 3 days work. Doesn't require the babysitting that eon does.