I think I need help with my Mac
Anonymous
12 Feb 2005 15:47:48 UTC
Topic 12511
(moderation:
> Have you tried quitting and restarting to see if it will switch. Also, you
> state that you are running boinc_4.19_ppcG4test. That was a Team MacNN beta
> testing compile, but don't know if it was an early one or a late one you have.
> Try going to http://members.dslextreme.com/~readerforum/forum_team/boinc.html
> and downloading the latest optimized BOINC compile (without test on the end).
> That is the latest optimized compile that has been stable.
>
> I just looked at your work units and show that work unit 32999 (result ID
> 1100632) did return at 16:21 UTC today (shortly after your post) Was that the
> problem work unit? It shows....
>
> [core_client_version]4.19[/core_client_version]
> [stderr_txt]
> No heartbeat from core client for 30.006097 sec - exiting
> detected finished Fstat file - skipping Fstat run 1
> Resuming computation at 3900/127620/127620
> APP DEBUG: Application caught signal 2
> detected finished Fstat file - skipping Fstat run 1
> Resuming computation at 17041/438743/438743
>
> [/stderr_txt]
>
> Looks like something strange was happening so it will have to leave it to the
> project team to sort that out.
Bruce Allen here (posting from my wife's E@h laptop/account!)
The messages above (from the application) are OK. The 'no heartbeat' message is a bug in the BOINC library that has resisted all of our efforts to track down and fix. However we have a workaround, so you can ignore that. The 'Resuming' message is a something that indicates a normal resumption from a checkpoint. The 'detected finished Fstat file' also indicates a reload from a mandatory checkpoint halfway through the running job. So all messages here look from the point of view of the einstein@home application as if all is as it should be.
Cheers,
Bruce
I think I need help with my Mac
)
> I didn't remove it from memory after all; I found that that is a general
> preference and it would also remove my CPDN from memory and I need to leave
> that in so I don't end up repeating a lot of calculating time.
>
> I tried resetting E@H to see if that would help, but no such luck. Following
> the reset, everything was going great until E@H took over again. After
> calculating for two hours, E@H says it's pausing and SETI begins to run, but
> when I check Activity Monitor, it says that E@H is still running and using
> half the CPU time, while SETI is using the other half. This is all on a
> single-CPU PowerBook G4. href="https://eah.studiodelta.us/%3Ca%20href%3D"http://mywebpages.comcast.net/freaner/Picture6.pdf">http://mywebpages.comcast.net/freaner/Picture6.pdf">Here[/url] is a screen
> shot of Activity Monitor showing E@H continuing to run instead of pausing like
> it says it is.
In your preferences, how often have you set the jobs to checkpoint? It should be frequently (at most, every few minutes).
Bruce
> No matter how small you set
)
> No matter how small you set the "write to disk every xx seconds" setting, if
> application cannot checkpoint that often, it will not checkpoint. CPDN cannot
> checkpoint often, it does so in my computer for example approx every 900
> seconds.
>
> This option won't help.
I thought that if the time was set very long (say 12 hours between disk writes or checkpoints) then the BOINC code that was switching from app A to app B might wait for the next checkpoint to be completed before shutting down. But this obviously isn't the problem, if the checkpointing is every fifteen minutes.
Bruce
> > I've finally had to
)
> > I've finally had to detach from the E@H project. No matter what I have
> tried,
> > E@H continues to say it has paused when it's supposed to switchover to
> another
> > app, but never does pause. I let it run all night, just in case it might
> be
> > needing some additional time for shutting down, but E@H apparently
> stayed
> > active and using CPU time the whole night.
> >
> > I'd really like to help with the E@H science, but I can't let the E@H
> client
> > monopolize the CPU. Bruce: do you know any other Mac PowerBook users who
> are
> > successfully running E@H with CPDN and SETI that I could correspond with
> via
> > email to try to find a solution?
> >
> > C
>
> C, I'm going to load up SETI to run with Einstein on my G5 and watch it for
> a few days, to see what happens. Maybe we can figure out if it is a BOINC
> client issue or something being triggered by Einstein.
Thank you. We'll look for any information you might provide.
Cheers,
Bruce