Now that the hubub over the new Boinc client has died down, I have a question that's been in the back of my mind for awhile. I like to ask questions here because I want to understand the program better, and I know some people here like answering questions.

Say Boinc tells me that it's deferring communication for 3 hours. If I turn a computer off for the night, and back on in the morning, does it count all that offtime as deferred time, or does it start the 3 hour countdown from when the computer comes back on?

Or, in other words, does Boinc count computer offtime in its deferred connection time, or is deferred connection time only counting down if the client is open?

Make sense?

One of my machines had "deferring connections 23 hours" when I shut it down near midnight, and something like "deferring connections 14 hours" when started this morning.

... and I've only seen those long times this week when the schedulers were down for a long time.


> Or, in other words, does Boinc count computer offtime in its deferred
> connection time, or is deferred connection time only counting down if the
> client is open?

> Now that the hubub over the new Boinc client has died down, I have a question
> that's been in the back of my mind for awhile. I like to ask questions here
> because I want to understand the program better, and I know some people here
> like answering questions.
>
> Say Boinc tells me that it's deferring communication for 3 hours. If I turn a
> computer off for the night, and back on in the morning, does it count all that
> offtime as deferred time, or does it start the 3 hour countdown from when the
> computer comes back on?
>
> Or, in other words, does Boinc count computer offtime in its deferred
> connection time, or is deferred connection time only counting down if the
> client is open?
>
> Make sense?
>
>
I dont know this for a fact but by example mine seems to keep a clock internally within the program. If I have an hour to go and completely close/exit boinc then load it back up sometime later it seems to pick up where it left off and I'll still have that hour wait. I'm hoping this is just an error on my perception and that it uses the computers system clock.

> Now that the hubub over the new Boinc client has died down, I have a question
> that's been in the back of my mind for awhile. I like to ask questions here
> because I want to understand the program better, and I know some people here
> like answering questions.
>
> Say Boinc tells me that it's deferring communication for 3 hours. If I turn a
> computer off for the night, and back on in the morning, does it count all that
> offtime as deferred time, or does it start the 3 hour countdown from when the
> computer comes back on?
>
> Or, in other words, does Boinc count computer offtime in its deferred
> connection time, or is deferred connection time only counting down if the
> client is open?
>
> Make sense?
>

I would say that BOINC is not "counting" but it uses your windows clock.
The next day you won't have to wait 3 hours again.

That's just a guess. But it make sense to me.


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

Make sense?

>As Near as I can tell on My machines the programs start anew and request new units, if they are ready then it downloads, (deferred connection time only counting down if the client is open)

But resets the clock so it does not retain wait time from shutdown to start up.....

Timbo
[url=http://www.boinc.dk/index.php?page=user_statistics&project=sah&userid=147175] </img>

---

> > Now that the hubub over the new Boinc client has died down, I have a
> question
> > that's been in the back of my mind for awhile. I like to ask questions
> here
> > because I want to understand the program better, and I know some people
> here
> > like answering questions.
> >
> > Say Boinc tells me that it's deferring communication for 3 hours. If I
> turn a
> > computer off for the night, and back on in the morning, does it count all
> that
> > offtime as deferred time, or does it start the 3 hour countdown from when
> the
> > computer comes back on?
> >
> > Or, in other words, does Boinc count computer offtime in its deferred
> > connection time, or is deferred connection time only counting down if
> the
> > client is open?
> >
> > Make sense?
> >
> >
> I dont know this for a fact but by example mine seems to keep a clock
> internally within the program. If I have an hour to go and completely
> close/exit boinc then load it back up sometime later it seems to pick up where
> it left off and I'll still have that hour wait. I'm hoping this is just an
> error on my perception and that it uses the computers system clock.
>

You are quite correct. I have shut down my PC overnight, when it had nothing to do but wait to contact the "scheduler", and the time started where it left off the night before. I hope this answers the original question. L8R....

---
<p>Rick A. - BOINCing right along now.... It can only get better! </p><p> Live Long and Prosper....</p>
<p>"There is no fate but that which we make for ourselves."

hubub? What is hubub?

allow me 2 things to add

the client is talking to us over the messages.

But even when the client talks deferring about ...1 Hour...1 week,
he is doing an other, internal "intervall",
from minimun intervall, about 1 minute to at highest,about 4 hours.

Why? I'dont now. It's an ucb client, ask ucb.

the other thing, all those nice "triggering" information are
saved on the clients pc.

and the third of the two thing is,
you start the client, the client will try to connect ucn as fast as it can AND overiding the local saved data with new actualized one.

Make sense?
to propper awnser this Q: pls first define what is sence

hope could sort a little


ric

When BOINC deffers communitcation it sets a variable to the date and time that it can connect after. So the clock is running weather or not the client is running since it just checks now > connect time. If the client is not deffering the variable will be zero. Sorry I don't remember the variable name off the top of my head and don't really have a means to dig it up from here.

JKeck

> You are quite correct. I have shut down my PC overnight, when it had nothing
> to do but wait to contact the "scheduler", and the time started where it left
> off the night before. I hope this answers the original question. L8R....

Well I am not so sure.

I've just made a test here. Message was
"no work from project - deffering communication with project for 1 hours"
I have quit boinc and restart it about 15 minutes later. message was
"Deferring communication with project for 44 minutes and 26 seconds"
It seems that the clock is tiking when boinc is not running. It's got
to be using system clock. If it's really the case then the clock is
also tiking during the night or when PC is off.

Next test I will quit boinc, advance windows clock, restart and lauch
boinc. I will post results later.


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

Thankyou Mark, please do the test.

I'm confused, how can the clock run when the PC is powered off
and the advance windows clock is not started

edit sorry missreaden

you wrote: advance windows clock
I read: advanced windows clock
sorry I made the confusion

> I'm confused, how can the clock run when the PC is powered off

It has a little battery on the Mother board that keeps it alive.
It's called RTC (Real Time Clock).

> > I'm confused, how can the clock run when the PC is powered off
>
> It has a little battery on the Mother board that keeps it alive.
> It's called RTC (Real Time Clock).
>

I am not a wiz at computers but I think the BOINC program just looks at the time that has lapsed from the time it was shut down until it was stared again and just reset its time....Hope some else can confrimn or refute my little assumption,,,,,,;o)

This 'SPACE' Rented.

The anonymity of the Internet, brings
forth, yet another EXPERT.

M7 Seti@h Berkeley's Staff Friends Club ©

> Next test I will quit boinc, advance windows clock, restart and lauch
> boinc. I will post results later.
>

OK I have made the test and I can confirm at 100% that BOINC doesn't have
a clock. When BOINC says deffering communication for 10 minutes it doesn't count. It will resume communication when those 10 minutes have elapsed on the system's clock. I've just fooled BOINC by changing windows time and it worked.

I am not a programer but I guess that it's the same for all programs. The only
time reference on the PC is provided by the RTC to Operating system and to
applications.

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

dear readers, If you want, take a closer look
into your client_state.xml file,

section min_rpc_time

update, watch the values, update, watch the values

> dear readers, If you want, take a closer look
> into your client_state.xml file,
>
> section min_rpc_time
>
> update, watch the values, update, watch the values

OK so the "watch the values" confirm that it refer to
system clock right ?

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

> OK so the "watch the values" confirm that it refer to
> system clock right ?

the reference basicaly, your are right.
The only possible "mesurement" is cpu usage and clocktime.

But now the ?? beginns, the client takes the sysclock as reference, but uses an internal, own "logic" for interpretation of delaying. normaly the cl delays with an linear increasing timing intervall.
Part of this we can observe in the msg windows. I'm "speaking" about the m$ client, not knowing if the mac side are used to see those "non linear" time delaying.

the minimal delay is about 1 minute, maximal delay about 4 h.

even if the client is offering us a delay of 18h 13 min 23 sec, no more time than 4 hour will go until next try.
also unknown, if the client takes the systime just while starting, or what I belief, the client is synchronizing time (and other information) on a frequent manner while.

em.. to be as clear as possible, we are talking about the Grafical User Interface part, not the client CLI part.

We in m$ we use both, the gui, "manages" the client, exception when the client
is started standalone.


Ric

BTM
Used with ferries, but indeed, not more.. 4me 50% change to have le mal de mer

> Used with ferries, but indeed, not more.. 4me 50% change to have le mal de
> mer

LOL You were feeding the fish...

>
> (Linked Object Library) You were feeding the fish...
not feeding, eating fish.

that was on the canarian islands, from gran canaria to teneriffe, 4h dark hours in my life, the ocean made this:. `·.¸¸.·´¯`·-

the town where my girl and I are,is called Basel, the town is "seperated" by a river, der Rhein, about 300-400m large. to cross the river we do have bridges and in Basel,5 small ferries, only for people transfer only motored by watercraft, a link attached.

for this distance, I'm not having the mal de la mer.
sorry not to be perfect..
`·.¸¸.·´¯`·-

ric

http://www.faehri.ch/

>
> the minimal delay is about 1 minute, maximal delay about 4 h.
>
> even if the client is offering us a delay of 18h 13 min 23 sec, no more time
> than 4 hour will go until next try.

The built-in exponential backoff should be between 1 minute & 4 hours, but some confusedly has reported this to be much longer. Not sure here if it retries after max 4 hours or not.

But atleast then you get "wrong major version" you'll get 24 hours deferral and it will count-down every hour till 0 hours before trying to connect again, so of course this is 24 hours. Running the CPDN-beta some weeks has also give this message from seti the last weeks. ;)

> >
> > the minimal delay is about 1 minute, maximal delay about 4 h.
> >
> > even if the client is offering us a delay of 18h 13 min 23 sec, no more
> time
> > than 4 hour will go until next try.
>
> The built-in exponential backoff should be between 1 minute & 4 hours, but
> some confusedly has reported this to be much longer. Not sure here if it
> retries after max 4 hours or not.
>
> But atleast then you get "wrong major version" you'll get 24 hours deferral
> and it will count-down every hour till 0 hours before trying to connect again,
> so of course this is 24 hours. Running the CPDN-beta some weeks has also give
> this message from seti the last weeks. ;)
>
There can be seperate conditions causing the backoff. Among them:
scheduler request - can be any timestamp sent by the server to your client.
scheduler fail - 1 min to 4 hours randomly increasing with each failure, after too many failures (currently set to 10) causes a master fetch.
master fetch - 1 min to 2 weeks randomly increasing with each failure.
file transfer - 1 min to ? (I think 4 hours but not sure) randomly increasing with each failure, will eventually give up (about 2 weeks).

This is not a complete list of types and limits to exponetial backoffs just some of the more common ones.

The backoffs work by inserting a timestamp into the min_rpc_time (and similar) tag after which the client can attempt that particular operation again. The logic is: if now > min_rpc_time allow. If no backoff is currently active min_rpc_time will equal zero. There is also seperate code that reports the approximate time until the backoff expires every hour.

John Keck -- BOINCing since 2002/12/08 --