I don't speak the language. What is a "WU"?

And how do I set up a buffer for it?

WU is Work Unit, which is each "chunk" of data that is downloaded, processed, and sent back.

To change it, go into General Preferences and change the "Maintain enough work for an additional
(Requires 5.10+ client.)" part to how ever many days worth of data you want to have downloaded.

I did have this set to the default, which was like .25, but just changed it to 5 days just in case this happens again :)

You can also increase the "Connect to the Internet every xx days" to something above default (.1) - I set it to 4.5 days - this works with all versions of the client, back to 5.4.8... (that I know of...)

---

.

Hello, from Albany, CA!...

I do agree with you 99.9% and I do feel bad that I can't send money.

I can control the amount of my spending on bills to the point of shutting down my PC when there is no work. But I do run other projects as well so I keep it on for them.

I never said I thought S@H was going to stay up 100% and I just picked the last 2 months as a sample.

Anyway, I am just venting so don't pay any mind to me.

---

Back on point, have there been changes at CentralCommand that would cause the following...

May 23, my computer gets oodles of ghosts, which are wu's that are assigned to my computer but were never downloaded. These ghosts are timing out and seem to be going away.

But last night (6/11-12) I now have oodles of new WU's downloaded to my computer. What is strange (apart from never usually getting so many because my queue is set to the minimum) is that three of them were started and now, including einstein, I have 4 wu's being time-sliced.

This seems like something is jerking my settings and/or there is a boinc bug seti is revealing (to me at least).

Comments?

Thank you to those of you who have answered my question.

If version 5.10 is a "B" model I shall leave well alone.

Why do they call it a "client"? A client is a customer.

---

Equal rights for equal responsibilities.
Equal opportunities for equal abilities.

@PhonAcq:

They enabled auto-resends finally. Since a ghost doesn't count as new work, if you had a bunch assigned, when you finally get through to the scheduler you will pick up the ones which are still eligible for crunching (ie, no quorum yet, the rest get canned), which can make it appear you exceeded your cache setting for that one contact session.

Depending on how fast the DL's arrived, BOINC might start one, then get another with a shorter deadline, and swap to that one and so forth. This ends up with a bunch running at the same time until the scheduling jam gets cleared from the ghost resend.

HTH,

Alinator

Thank you to those of you who have answered my question.

If version 5.10 is a "B" model I shall leave well alone.

Why do they call it a "client"? A client is a customer.

The part of BOINC you run on your host is called the 'client' for the exact reason you mentioned. It's the 'customer' of the various projects you attach to. It's roots are in the client/server model the BOINC system uses.

The part that actually does the work is called the science application for obvious reasons. ;-)

HTH,

Alinator

I think as people get tired of trying to get work or upload/download they will look to other projects.

Like the infomercial... SET IT and FORGET IT!

That is one of the great BEAUTIES of BOINC.
Who cares if there is no work today or tomorrow, when there is work my computers will HELP on a non-interference basis.

I have 2 remote machines I've setup & NEVER been back to! They do what they can when they can. One machine is turned of for months at a time, but when it is on & not being used it helps out. If the operator leaves the machine off for a week, oh well. That is the project-side beauty of the BOINC architecture: if I don't return a result in a 'reasonable' time, the WU can be sent to others with no significant harm to the project or me (ok, well I don't get some credits, it's not like I'm loosing money or even MY TIME!).


I try to attach each computer to at LEAST TWO (2) projects with drastically different SHARES set (i.e. 100 for SETI, 10 for #2, 1 share for #3 project). If project #1 runs out of work or has server problems, project #2 can pick up the slack. Same for project #3.


Are you 100% busy, every single second at work?
CPUs need coffee breaks and vacations too ;)

SET IT & FORGET IT !!

---

SETI@home since 1999
"Set it, and Forget it!"

@PhonAcq:

They enabled auto-resends finally. Since a ghost doesn't count as new work, if you had a bunch assigned, when you finally get through to the scheduler you will pick up the ones which are still eligible for crunching (ie, no quorum yet, the rest get canned), which can make it appear you exceeded your cache setting for that one contact session.

Depending on how fast the DL's arrived, BOINC might start one, then get another with a shorter deadline, and swap to that one and so forth. This ends up with a bunch running at the same time until the scheduling jam gets cleared from the ghost resend.

HTH,

Alinator

Thanks. I noted that the 'new' wu's have much later deadlines, if that matters. I've been reducing my resource allocation to seti due to all this backroom chaos; it will be interesting to see if all these wu's actually meet the new deadline, given that einstein is taking a time slice.

HTH ?= {hit the head?, hollar to hilda?, honor thy husband?}

HTH ?= {hit the head?, hollar to hilda?, honor thy husband?}

Hope That Helps

---

I just made a little graphic to illustrate the boinc client/applications/wu model...

AFAIK (as far as I know), HTH is "happy to help". YMMV (your mileage may vary)

AFAIK (as far as I know), HTH is "happy to help". YMMV (your mileage may vary)

WTF?

AFAIK (as far as I know), HTH is "happy to help". YMMV (your mileage may vary)

WTF?

Fondly quoting MSattler: "Where's the Fish?"

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

Seems better to be without for a week instead of wasting time trying to get work uploaded or downloaded and reported or whatever.

Considering the squealing that goes on when the system is down even if only for a few minutes, better to have it up & provide some (even if extremely sporadic) work than none at all.
As it is, many of the problems tend to be load related.
So while things might be running nicely, once the system gets loaded up again the problems make themselves known. Such things are always so much fun to track down. Tweak & restart, tweak & restart. And so it goes.

---

Grant
Darwin NT

And I always thought WTF meant Wand To Flamenco!

AFAIK (as far as I know), HTH is "happy to help". YMMV (your mileage may vary)

WTF?

Fondly quoting MSattler: "Where's the Fish?"

---

Kenn

What is left unsaid is neither heard, nor heeded.
Ce qui est laissé inexprimé ni n'est entendu, ni est observé.

Jocelyn has been stoped, and we have 'The database server is not accessible' on Database status...

---

Jocelyn being down is no big surprise, given what Matt said here.

As far as the database status message goes, ignore it. Sidious is the one which really counts and the monitor must not have been updated to reflect he's the primary BOINC DB server at the moment.

Alinator

I think the part of the point that some are missing that Ricky is trying to make is that the project is about participation. The more participation the better. However, with the lackluster performance as of late, volunteers are switching to other projects (I have). That is sort of defeating part of the reason for having the project in the first place. Volunteers. Don't try to read too much into it...

---

"we're once again finding ourselves frustrated with a random and pointless problem."

I thought they were all that way....

Well, welcome to my world anyway...

Thanks for the effort!

---

We are all Alien. It just depends on which planet you are standing on.

I just made a little graphic to illustrate the boinc client/applications/wu model...

This graphics is only slightly wrong. The central box should be BOINC or BOINC client rather than BOINC manager. There's another box missing right next to the box with boinc_cmd which is for BOINC manager and has the same interaction with the rest of boxes as boinc_cmd does.
The only difference between BOINC manager and boinc_cmd is graphics-ness of the former.

---

Metod ...