We'll just about get the system working, and then we'll roll out a major and incompatible upgrade to the client. We'll give only 24hours notice so nobody has time to report the work they've done or clear their caches. We have yet another bug, so we'll keep the scheduler off line just to make sure that work can't be completed. Then, we'll roll out the client on a Friday afternoon so that when it all goes wrong there's nobody about to fix it...

Oh - and just to make sure that the disaster really does happen we won't give any official details of the effect of the upgrade until the last minute.

Well, in the beta nearly all releases was on Friday... Still it's 11:36 Berkeley-time, so they should probably have some hours to monitor the system.

As for giving notice, look on the message from 25. August... so looks like 2 days for me...

Also, it has been impossible to download work the last two days so most haven't got anything.

> We'll just about get the system working, and then we'll roll out a major and
> incompatible upgrade to the client. We'll give only 24hours notice so nobody
> has time to report the work they've done or clear their caches. We have yet
> another bug, so we'll keep the scheduler off line just to make sure that work
> can't be completed. Then, we'll roll out the client on a Friday afternoon so
> that when it all goes wrong there's nobody about to fix it...
>
> Oh - and just to make sure that the disaster really does happen we won't give
> any official details of the effect of the upgrade until the last minute.

Was this an official message? *lol*


S@h Berkeley's Staff Friends Club © member

Ingleside, check the message for 25th again - it says 'tomorrow'. 'Tomorrow' didn't happen, but it made no difference because the scheduler was off line anyway due to a bug. Let's not split hairs...

BOINC: SETI@home, Astropulse, CPDN, Predictor@home, etc. etc.

An ENORMOUS INTERNATIONAL GIANT

always without money and in this moment

based on only f-o-u-r pillars of "manpower":

Dave, Rom, Matt & Jeff,

the Fab Four of Distributed Computing.


[/url]

If 24 hours are not enough to flush cashes, MikeW, how much would be? A week? A month?

> Ingleside, check the message for 25th again - it says 'tomorrow'. 'Tomorrow'
> didn't happen, but it made no difference because the scheduler was off line
> anyway due to a bug. Let's not split hairs...
>

Hmm, the scheduling-server said "no work available" earlier today so has been up atleast partly, but hadn't any results to return so maybe this wasn't working...

>> If 24 hours are not enough to flush cashes, MikeW, how much would be? A week? A month?

It's taken me two days to crunch the remaining work in my cache, not that I have been able to report it. Others may have more. I'd have thought a week would have been a good estimate.

AND the SETI team could have used the time to work through any outstanding server side problems, maybe posted some decent info on the upgrade and perhaps even made the new client available...

> >> If 24 hours are not enough to flush cashes, MikeW, how much would be?
> A week? A month?
>
> It's taken me two days to crunch the remaining work in my cache, not that I
> have been able to report it. Others may have more. I'd have thought a week
> would have been a good estimate.
>
> AND the SETI team could have used the time to work through any outstanding
> server side problems, maybe posted some decent info on the upgrade and perhaps
> even made the new client available...
>
You're probably right - a week is better. But they (Berkeley) simply don't have this week. Many crunchers are affected now, no doubt - but I think a better way is to accept the tings as they are. :)

> You're probably right - a week is better. But they (Berkeley) simply don't
> have this week. Many crunchers are affected now, no doubt - but I think a
> better way is to accept the tings as they are. :)


They would have had the week if they had prepared the transition one
week before....... Don't you think so ?

-.-. --.- -.. -..- . - .-.-. -.- --... ...--

I've given my 4 puters the appropriate enema (Reset) and I'm waiting....... ever so patiently

>
> They would have had the week if they had prepared the transition one
> week before....... Don't you think so ?
>
> -.-. --.- -.. -..- . - .-.-. -.- --... ...--


I think all the story with the issues last 3 weeks was an emergency situation.
If you check the history you would see:

"August 5, 2004
A misconfigured scheduling server ran for about an hour this morning, misinforming users that they had to upgrade to version 4 of the BOINC software. This message was in error - please remain with version 3."

That's 3 weeks ago. A "misconfigured scheduling server" sent messages that was supposed to be sent today. They (Berkeley) had most of the things ready 3 weeks ago.
But then happened a HW strike and everything went wrong. Can you guarantee for any of the HW parts in your machines? I don't think so. Nobody can, in fact. That's why I call this "an emergency situation".
And what is more - it is remarkable that they (Berkeley) succeded to overcome most of the things and to have brand new HW (no doubt such transition was planned but for uncertain future) functioning.
That's why I don't thing it is fair to blame them (Berkeley) in so many posts here and to try not to see their (Berkeley) enormous effort. Don't you think so?

The problem is that the announcement says that any work done on the 3.x ckient will not be accepted after the upgrade to 4.x. This is the first I have heard of this. Every time the question was asked "can we continue to run 3.x till our cache is empty?" The answer was I think so...this is the way it worked in beta. Why did not somebody from the dev team come on the board and say that after the transition we will no onger accept credits from the 3.x client...and if this is truly the case, Why are they still sending out credits to the 3.x client? This makes no sense.

Don

Hi everbody.

Everytime something happens that includes some kind of "disaster" that would mean loss of credits or processed WU, we can always find some posts of very unsporty nature trying to blame or bring down someone.

It almost seems like where talking about a life or death situation for some ppl sometimes.

Where in here by choice, fun, the project itself and what it means and a little competition amongst ourselves no doubt.

But there's no reason to get all worked up everytime a "little" setback occurs.

So when this kind of thing happens, the best we have to do is to have a drink with some friends, especially on a Friday night and wait for further developments.

Keep it cool and running ...

> Hi everbody.
>

> So when this kind of thing happens, the best we have to do is to have a drink
> with some friends, especially on a Friday night and wait for further
> developments.
>
> Keep it cool and running ...
>
>
You're right - today is friday :))

I just looked at my machine that is still running 3.2.

All of the completed work units have been uploaded.

It has a couple of new work units that it won't get to -- those I'm sure will be reassigned after 4.0 is released.

It's about half way through one that will likely be lost.

That doesn't sound like a really big deal, especially since the machine would have been idle otherwise.

We're not really supposed to be buying machines and setting them up just to run SETI/BOINC. There are more wasted clock cycles than work -- at least until a few more projects get rolling.

-- Lynn

> Hi everbody.
>
> Everytime something happens that includes some kind of "disaster" that would
> mean loss of credits or processed WU, we can always find some posts of very
> unsporty nature trying to blame or bring down someone.
>
> It almost seems like where talking about a life or death situation for some
> ppl sometimes.
>
> Where in here by choice, fun, the project itself and what it means and a
> little competition amongst ourselves no doubt.
>
> But there's no reason to get all worked up everytime a "little" setback
> occurs.
>
> So when this kind of thing happens, the best we have to do is to have a drink
> with some friends, especially on a Friday night and wait for further
> developments.
>
> Keep it cool and running ...
>
>
When someone in control and in-the-know doesn't spend five minutes to update the news page, well, I sure couldn't justify that and don't know how they can/do. Just because we volunteer our time, equipment, and energy (to run the computers) that doesn't mean we shouldn't be informed so that we don't waste or time and energy.

> The problem is that the announcement says that any work done on the 3.x ckient
> will not be accepted after the upgrade to 4.x. This is the first I have heard
> of this. Every time the question was asked "can we continue to run 3.x till
> our cache is empty?" The answer was I think so...this is the way it worked in
> beta. Why did not somebody from the dev team come on the board and say that
> after the transition we will no onger accept credits from the 3.x client...and
> if this is truly the case, Why are they still sending out credits to the 3.x
> client? This makes no sense.
>
> Don
>
>
That is not how it worked in beta. The major versions were completely incompatible, you couldn't even get preferences. With the change to version 3.xx there was supposed to be a change that would enable some communications sometimes, but expect that once the servers change over all outstanding work will be void.

John Keck
BOINCing since 2002/12/08