Unknown application 'hsgamma_FGRPB1G' ? It was 'known' before I did a reset..

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 352
Credit: 512,495,766
RAC: 2,861,133
Topic 224543

 

I have found this in my event log, and I don't understand it.  I did a reset of E@H in BOINC Manager and now this shows up when earlier it was just fine.


01/19/21 9:58:48 AM | Einstein@Home | Your app_config.xml file refers to an unknown application 'hsgamma_FGRPB1G'.  Known applications: 'einstein_O2MD1'


Could someone explain to me what happened?

And another thing I find unusual is the time it is taking for an 'update' in E@H, approximately 9 hrs.  M@H & U@H are normal at about 10 seconds for Universe but Milkyway is repeating itself at about 1:30 seconds to update over and over.  What gives?

George

Ian&Steve C.
Ian&Steve C.
Joined: 19 Jan 20
Posts: 853
Credit: 4,613,187,573
RAC: 31,529,995

do you have this host set to

do you have this host set to only get GW tasks?

the GR tasks wont be "known" until you actually receive some after the reset. BOINC doesnt find any reference in your client_state file, that's why it thinks it's unknown. if you've received some GR GPU tasks since the reset, then restart BOINC and you shouldnt see that message anymore.

_____________________________________________

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 1,304
Credit: 2,147,377,017
RAC: 11,756,565

You haven't received any work

You haven't received any work yet to match the application. Just wait for new work.

The server contact time is set by each project and is not controllable by the end-user at least with Windows.

You do have the option to run our Linux client that can control that parameter as a GPUUG team member.

I think that a lot of projects are just "too busy" with host communications.  What is the point to constantly banging the scheduler every 11 seconds at Universe when the tasks take hours to run.

Or the same at MW which tries to connect every 91 seconds.

 

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 352
Credit: 512,495,766
RAC: 2,861,133

Ian&Steve C. wrote: do you

Ian&Steve C. wrote:

do you have this host set to only get GW tasks?

the GR tasks wont be "known" until you actually receive some after the reset. BOINC doesnt find any reference in your client_state file, that's why it thinks it's unknown. if you've received some GR GPU tasks since the reset, then restart BOINC and you shouldnt see that message anymore.

In the E@H website project preferences I have only "Gamma-ray pulsar binary search #1 (GPU)" checked off.

In my 'app_config.xml' file I have '<name>hsgamma_FGRPB1G</name>' listed as the only option.

So NO, I do not have any GW tasks set in this host.

What you say about the GR tasks not being "known" until I actually receive some makes a lot of sense.

Thank you Ian&Steve.  I was impatient as usual, but it makes more sense to me now.

Do you have any explanation as to why I'm getting about 9 hours delay when I update?

 

George

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 1,304
Credit: 2,147,377,017
RAC: 11,756,565

Since you don't have any work

Since you don't have any work yet, the normal 60 second Einstein connect time is pushed out, longer and longer by the scheduler at each connect

You can always manually update the project using the Manager's Projects tab to force a connection.

 

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 352
Credit: 512,495,766
RAC: 2,861,133

Keith Myers wrote: You

Keith Myers wrote:

You haven't received any work yet to match the application. Just wait for new work.

The server contact time is set by each project and is not controllable by the end-user at least with Windows.

You do have the option to run our Linux client that can control that parameter as a GPUUG team member.

I think that a lot of projects are just "too busy" with host communications.  What is the point to constantly banging the scheduler every 11 seconds at Universe when the tasks take hours to run.

Or the same at MW which tries to connect every 91 seconds.

Sorry, I didn't see your post until I responded to Ian&Steve's post.

Thanks Keith for your wisdom and for giving me another reality check.  I'm presently mulling over, seriously, about switching over to Linux on my 3950X machine.  It may happen in a few days.

George

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 352
Credit: 512,495,766
RAC: 2,861,133

Keith Myers wrote: You can

Keith Myers wrote:

You can always manually update the project using the Manager's Projects tab to force a connection.

If I understand you correctly, in BOINC Manager under the Project tab, when I highlight E@H and then select update it should automatically update?  Do I have this correct?

This is when I get about 9 hours waiting.

George

archae86
archae86
Joined: 6 Dec 05
Posts: 3,001
Credit: 4,770,180,997
RAC: 3,602,449

George wrote: Keith Myers

George wrote:

Keith Myers wrote:

You can always manually update the project using the Manager's Projects tab to force a connection.

If I understand you correctly, in BOINC Manager under the Project tab, when I highlight E@H and then select update it should automatically update?  Do I have this correct?

This is when I get about 9 hours waiting.

The most recent contact log information is something you'll want to consult in this sort of occasion.  It is linked from the details page for your computer with the link display text showing the time/date of most recent connection.

Some important lines from the most recent contact log for your Windows machine just before I started typing:

 stopping work search - daily quota exceeded (136>=41)
Daily result quota 41 exceeded for host 12851564
No work is available for Gamma-ray pulsar binary search #1 on GPUs
(reached daily quota of 41 tasks)

The number 41 is a big clue here.  Your machine does not have that low a daily quota in the natural state, so one can infer you generated errors on that machine today, lots of them.

Consulting the "errors" link on the tasks list for that machine, the indication is that you aborted a great many GW CPU tasks at 2:54 a.m. UTC on January 19 (UTC).

So at this point, no matter how often you mash the update button, you'll get no more tasks until your quota is topped up.

Upping your quota happens two ways: non-error task return, and the passing of the current "day'.

The benefit from returning tasks is very strong, so if you can get some of the few remaining GW CPU tasks to complete and return them, just possibly you might get your quota up past 136 for today, which would allow new task download.  I think it doubles your quota for each returned task, so just two might be enough.

There is a sticky point regarding the "day".  Your deferral reads as if the end of the day for this purpose is midnight UTC.  However, in actual practice, requests at the moment that deferral expires are often denied, with language implying that the real "day" boundary for this purpose in this case is not midnight UTC.  But you can try every hour or so, and eventually, all will come right.

The bigger question is, how is it that you continued to download large numbers of GW CPU tasks you chose not to run?  Could it be that the preferences you are modified are not the ones that apply to the location (aka venue) actually in use for your machine?

 

 

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 352
Credit: 512,495,766
RAC: 2,861,133

Quote:archae86 wrote: If I

Quote:

archae86 wrote:

If I understand you correctly, in BOINC Manager under the Project tab, when I highlight E@H and then select update it should automatically update?  Do I have this correct?

This is when I get about 9 hours waiting.

The most recent contact log information is something you'll want to consult in this sort of occasion.  It is linked from the details page for your computer with the link display text showing the time/date of most recent connection.

Some important lines from the most recent contact log for your Windows machine just before I started typing:

 stopping work search - daily quota exceeded (136>=41)
Daily result quota 41 exceeded for host 12851564
No work is available for Gamma-ray pulsar binary search #1 on GPUs
(reached daily quota of 41 tasks)

The number 41 is a big clue here.  Your machine does not have that low a daily quota in the natural state, so one can infer you generated errors on that machine today, lots of them.

Consulting the "errors" link on the tasks list for that machine, the indication is that you aborted a great many GW CPU tasks at 2:54 a.m. UTC on January 19 (UTC).

So at this point, no matter how often you mash the update button, you'll get no more tasks until your quota is topped up.

Upping your quota happens two ways: non-error task return, and the passing of the current "day'.

The benefit from returning tasks is very strong, so if you can get some of the few remaining GW CPU tasks to complete and return them, just possibly you might get your quota up past 136 for today, which would allow new task download.  I think it doubles your quota for each returned task, so just two might be enough.

Yes, I surmised that when I checked my details page after resetting my project.

archae86 wrote:

There is a sticky point regarding the "day".  Your deferral reads as if the end of the day for this purpose is midnight UTC.  However, in actual practice, requests at the moment that deferral expires are often denied, with language implying that the real "day" boundary for this purpose in this case is not midnight UTC.  But you can try every hour or so, and eventually, all will come right.

The bigger question is, how is it that you continued to download large numbers of GW CPU tasks you chose not to run?  Could it be that the preferences you are modified are not the ones that apply to the location (aka venue) actually in use for your machine?

This I do not know.  I can tell you that for my 3950X machine I am requesting only <name>hsgamma_FGRPB1G</name> through my app_config.xml file for E@H and Gamma-ray pulsar binary search #1 (GPU) on my E@H preferences web page.

The current downloads of GW CPU tasks are from after resetting my computer.

 

 

George

Keith Myers
Keith Myers
Joined: 11 Feb 11
Posts: 1,304
Credit: 2,147,377,017
RAC: 11,756,565

You have Four Project

You have Four Project Preferences pages. One for each location/venue.

Are you sure you are modifying the correct venue that the host is assigned?

I'm not sure if resetting a host causes it to fall back to Generic location.  It might do that.

And if the host was on a different venue like Home, School or Work, your project app preferences you think are applying may not be valid.

 

 

GWGeorge007
GWGeorge007
Joined: 8 Jan 18
Posts: 352
Credit: 512,495,766
RAC: 2,861,133

Keith Myers wrote: You have

Keith Myers wrote:

You have Four Project Preferences pages. One for each location/venue.

Are you sure you are modifying the correct venue that the host is assigned?

I'm not sure if resetting a host causes it to fall back to Generic location.  It might do that.

And if the host was on a different venue like Home, School or Work, your project app preferences you think are applying may not be valid.

Yep, I just checked and my preferences page was still set on 'home'.  I checked 'generic' anyway and I found it was still checked off for almost all, so I checked that it was the same as my 'home' venue.  I then went and reset my E@H and we'll wait and see what happens.

As always, thank you AGAIN for bailing me out.    ;*)

 

George

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.