Eon server is coming back!!!!!

[Bok]

I just got word from Graeme :)

There looks to be a few hiccups at first so don't start em all up right away.

Project update coming soon....

Bok

:banana:

[FoBoT]

:drums:

[offamychain]

Yeah, I noticed that about 30 minutes ago on the one I left on it. It's done 4 WU's & 2 of those have been the "v4NetSend" errors, but at least it's back. It'll straigten out soon.

I'm adding memory to a couple machines later 2nite, so will switch them, + all the Linux's, over then.

Maybe our competition hasn't noticed this quite yet (but will when they read this post:cool: :cool: :cool: ), so it may be a good time to get a jump-start.

[IronBits]

:elephant: :whip:

[BrianM]

OOOOOOOOOOOOOOOO Boy Here We Go!!!!!!!:drums: :elephant: :thumbs: :|party|:

[Helix_Von_Smelix]

Sorry guy's. Switched over my boxes, now i have net send errors. on each box for each 10mins of work i then get 20-30 mins of downtime. switching back to other DC's.

Will set up to run with LM, GOD how does anyone run eOn with the server soooooo sloooow. Hasn't got any better as the day has gone on.

[Paratima]

Still getting a fair percentage of "v3NetSend Failed" errors.

I hope that just means, "Had an error, but results got thru on a retry & everything's cool".

Hope it doesn't mean "Dropped your last bucketful of points, too bad".

Part of what I have been loudly wishing for, on this project, is some documentation about stuff like this. Oh, well... POR

[Paratima]

Ummm, if option 2, mentioned by me above, is true,


"Dropped your last bucketful of points, too bad".then we're wasting a LOT of firepower. I'm now getting these v3net thingies almost all the time... :mad:

[Helix_Von_Smelix]

Looks to be ok at the moment. Will check my points at the next updated to see if the points are lost.

[willy1]

Is there ANY POSSIBILITY of getting answers to these questions from the project people? Have they vanished?

[Bok]

Yes there is,

please be patient though, as I mentioned last night, Graeme is in the middle of moving across country.

Bok

[Paratima]

Originally posted by Bok
Graeme is in the middle of moving across country. Which direction? :D

[Bok]

I would say SSE ;) :jester:

[Helix_Von_Smelix]

Checked error log on 8 cpu's i have 75 netsend errors. when they have at least one upload on each pc i get only 14 points. So work is being trashed, and cycles waisted.

I now know how to stop it happening again.
This should have been sorted out before the gauntlet

[Paratima]

Long as we're saving up questions for graeme, here's one I just got:

Dimer DoAssignment
Dimer: Start
Image Done: U: 0.622092 FC: 1790 TS: 293
Dimer: Done
HSize: 102
Bad Pref
Dimer UndoAssignment

What's a Bad Pref? Does my computation still count? Does a Bad Pref advance the science at all? What's the % of Bad Prefs to good ones?

Question #2 is kinda important to me, for some reason.

[offamychain]

Guys, can we PLEASE concentrate on the main goal of this whole Gauntlet...which was to take back 2nd place? There is only 1 way to do that, & it is to crunch. I have too many errors in my log to count, always have, always will. Bok does too, + every other individual running Eon.

It might be nice to know WHY, but I don't concern myself with that greatly because I really don't have time to baby-sit a dozen machines that are doing what they were built to do. If you're getting errors, everyone else is to, including the very team we're trying to pass.

They're letting their Eon systems run, like they want to, & not constantly pestering them to see how many of the last 10 WU's had "v3NetSend Failed" on them. Until we do the same, we WILL NOT succeed in passing them---bottom line---in ANY project. Might as well forget it, give it to 'em, & go play with something else.

ALL DC-projects, without exception, have MAJOR problems at times, sometimes often. If you're wanting something that will run forever on your computer & never screw up, you need to find another hobby, DC ain't the one for you.

I understand your concerns, but all these same questions have been asked, & answered, over & over again in the last week or so. Bok has been nice enough to contact Graeme & has posted the results of their conversation at least 3 or 4 time since yesterday. Until he has time to do something, let's not run this into the ground.

We started the Gauntlet as a team, it was not one individuals decision, & it had ONE specific goal in mind. That goal IS do-able, but it will take the team to do it. Are we gonna go down this same road on every Gauntlet we attempt?? I hope not. I've not been in ANY Rally or Gauntlet that didn't have problems.

Yes, I am partial to Eon. I had BIG problems with it at first, till I learned to leave it alone & let it work. But I also crunch whatever this team needs me to do, & I do it without question. If the thing has problems, I try to fix them, if it doesn't appear that I can, I accept that & just give the team whatever points I can make.

Let's come together on this thing, accomplish what we set out to do in the 1st place, & then move on to the next one. With that said, let's get done to what we do best--crunch data.

[Helix_Von_Smelix]

"goal do-able", look at the stats.


"questions answered", not seen that.

[rshepard]

I figure I'll let it run-- When the server is down for reloading, my backup project kicks in nicely ( except for Monday when its server was down as well :bang: )
And as for the work that sometimes gets lost due to the v3Send error -- well, maybe that will get sorted after the server move, maybe not--- but it's a level playing field either way. I refuse to get to worked up about any project; I do this for fun more than anything, and if/when this project stops being fun to me, I'll swing the boxes to something else.
Keep 'em crunchin' people, whatever you run :thumbs:

[Paratima]

offamychain: You may misunderstand the role of questioning here. We are looking for ways to improve the whole project's performance, as well as our individual and team performance. The project admins may or may not know about these problems. They may or may not have the resources or motivation to fix them, but surely if they don't know about them, the problems won't fix themselves. That's at least one reason we are so voluble.

None of us wants our work wasted, even forgetting the points and the competition. Bottom line, we don't want our boxen burning gigahertz to no effect. Each of us has his/her own threshold. If I find that some critical percentage of my work is being trashed, I'm out of here. The crit % varies by individual. The more it looks like approaching my personal crit %, the more I'll want explanations. At some point, I'll just say fsck* it!

Finally, I won't EVER stop asking questions until the day I go into horizontal off-line storage. Want me to run your project? Better explain what it does. Is it not running optimally? Tell me why not and your intentions regarding fixing it. Nada? Adios!

*fsck: file system check - Unix

[willy1]

and I want someone to quit preaching to the group on HOW to run the project. Most of us have been involved with DC for a very long time.

If I want to check every single machine 50 times a day, I will! And I damned well want to know why 3 out of 5 WUs can't get sent to the server! And, why the client wasn't designed to save them up for later transmittal instead of tossing them in the bit bucket!

Is that too much to ask?

[willy1]

Is the "RecvIdaMessage failed" in the client.log file the same error as the "v3 Net Send failed" error? I started a new log this morning, and already have hundreds !!! of those errors.

Is this a server workload problem? Can anyone answer that?

Or is it the network at the Univ of WA problem?

Or not enough work loaded in the server?

Or all of the above?

Is there a simple fix to the client (or server ?) throwing away the work when the NetSend error occurs?

Some people are obviously in contact with this mysterious 'Graeme' person. Does he need help with this thing? If he can answer emails he should be able to answer these questions clearly and succinctly and publicly!

[Bok]

Originally posted by willy1
Is the "RecvIdaMessage failed" in the client.log file the same error as the "v3 Net Send failed" error? I started a new log this morning, and already have hundreds !!! of those errors.


Yes it is, IMO



Is this a server workload problem? Can anyone answer that?

Or is it the network at the Univ of WA problem?

Or not enough work loaded in the server?

Or all of the above?


Possibly, we don't know and won't know until Graeme has more time to debug once he is in his new job



Is there a simple fix to the client (or server ?) throwing away the work when the NetSend error occurs?


If there was we'd probably have it. The source code is available if you'd like to look.



Some people are obviously in contact with this mysterious 'Graeme' person. Does he need help with this thing? If he can answer emails he should be able to answer these questions clearly and succinctly and publicly!

Yes he does need help, someone to rewrite the scripts which control the project and a DB to manage the userbase etc. Once in his new job, he will have undergraduates who will help with this. In the meantime I'll be helping in any way I can.

Bok

[rshepard]

now here's something odd-- I was looking at the outfile on my Linux box, and where it has the v3netsend error, it shows a non-zero bytes_sent value for each one. As I recall, the Windows clients I have running all have shown a bytes_returned value of zero on the same error. :confused:
Not sure what that means, if anything, but thought I'd note it. Speculation: the Linux client is not "faster", but rather is not losing the results on the v3 error, while the Windows client does.

[offamychain]

Just to clarify what I was attempting to do, I wasn't trying to "preach" to anyone about how to run this project. I believe the team decided to have an Eon gauntlet to try & take back 2nd place. No 1 person can do this, it takes a team.

Some of us are better at certain projects than others simply because maybe they're our personal "pet" projects. I believe I'm decent at Eon & Chess. Others are having trouble with Eon, so I've throwed every suggestion that I know of out there to try to get everyone's rigs going.

But that's all they are...suggestions. MY Eon is running right & has been for several weeks. Maybe I'm doing something right, so I don't mind offering help. If you want to do it your own way, I could care less..that's you. I joined this team with one goal in mind...to help it get as high in each project that I run as I can.

If I've got some hints that may help someone on a certain project, I don't mind shareing. Likewise, I'll run ANY project this team decides to concentrate on. If I don't know how, I'll do it the way someone who DOES know how does it.

Everything about Eon has been said in this forum the past week or so. If ANYONE actually knew how to stop these errors, make the server run all the time, etc., etc., they would have shared it. But we don't, so we work with what we have.

If my suggestions, hints, & help offended anyone, as they obviously did, I do apologize, I didn't mean it that way, & was trying to help accomplish what we set out to do. With that in mind, I've probaly already said to much, so will just run Eon, with the other members of the team who decide to participate in the Gauntlet, until we either take 2nd, or decide not to.